Google ads Interview Questions

Note: Lead with numbers. A paid search interview is fundamentally about whether you can be trusted with a budget, and vague answers suggest you have never owned one.

Structure it as:

• The business and the objective. Lead generation, e-commerce revenue, app installs, or brand awareness — these need completely different structures and metrics, and naming the right one matters.
• The budget and scale. Monthly spend, number of campaigns, and whether you owned the account or supported someone who did. Be honest about this.
• What you actually changed. Restructuring campaigns, moving to a smart bidding strategy, rewriting ad copy, fixing conversion tracking, adding negative keywords, or improving landing pages.
• The result, with before and after. Cost per acquisition, ROAS, conversion rate, or impression share — plus the timeframe.

If a campaign failed, say so and explain what you learned. Everyone who has managed spend has had one, and pretending otherwise is transparent.

Translate platform metrics into business outcomes, and lead with the answer rather than the data.

• Start with the money question. "We spent 4 lakh and generated 62 qualified leads at 6,450 each, against a target of 8,000" is the headline. Impressions, CTR, and Quality Score are supporting evidence, not the story.
• Use their language. Cost per acquisition rather than CPA on first use; return on ad spend explained as "we earned 4.20 for every rupee spent".
• Show the trend, not the snapshot. A single week is noise. Comparing month over month, and flagging seasonality, prevents panic over normal fluctuation.
• Be direct about what is not working and what you propose to do. Clients lose confidence when bad news only surfaces after they find it themselves.

Note: The strongest thing you can demonstrate here is managing expectations early — explaining that a new campaign needs a learning period, or that a smart bidding change will look worse for two weeks before it improves. That conversation is much easier before the dip than during it.

Show a diagnostic order rather than a list of tactics — that is what separates someone who has run accounts from someone who has read about them.

Check in this sequence:

• Is the tracking correct? Before concluding a campaign has failed, confirm conversions are actually being recorded. A broken tag makes a profitable campaign look dead, and this is the most common cause of a sudden collapse.
• What changed, and when? The change history shows edits; a sharp break points to a change, a gradual decline points to competition or seasonality.
• Where in the funnel is the loss? Low impressions is a budget, bid, or targeting problem. Good impressions with low CTR is an ad copy or relevance problem. Good clicks with no conversions is a landing page, offer, or audience-quality problem. This one distinction directs everything that follows.
• Check the search terms report. Irrelevant queries draining budget is extremely common with broad match.

Then fix, and change one significant thing at a time so you can attribute the result.

Answer with a system, because the real skill here is not doing everything — it is knowing what to ignore.

• Prioritise by spend at risk. A campaign spending 3 lakh a month deserves attention before one spending 15,000, regardless of which is more interesting.
• Separate the urgent from the routine. Broken conversion tracking, a disapproved ad, or a budget being exhausted by midday are same-day issues. Ad copy testing and keyword expansion are scheduled work.
• Use automated rules and alerts so problems find you rather than depending on you checking. Alerts on cost anomalies, conversion drops, and disapprovals cover most emergencies.
• Batch similar work. Doing search term reviews across every account in one session is faster than context-switching per account.
• Resist over-optimising. Daily bid changes on a smart bidding campaign actively harm it by resetting the learning period. Knowing when to leave something alone is a real skill.

Note: Mentioning a weekly and monthly cadence — what you check daily, weekly, and monthly — gives a concrete, credible answer.

How you keep up: the official Google Ads blog and release notes, industry publications such as Search Engine Land, and the practitioner community where people report what a feature actually does rather than what the announcement claims. Google's own guidance always favours automation and broader targeting, so it is worth reading critically.

How you decide what to test:

• Does it address a problem this account has? A new asset type is irrelevant if your bottleneck is landing page conversion rate.
• Can you test it without risking the account? Test on a smaller campaign, or a portion of budget, before rolling out.
• Can you measure the result cleanly? If the change coincides with a seasonal peak you will learn nothing. Use experiments and drafts where possible so you get a genuine comparison rather than a before-and-after.

Note: Being able to name something Google pushed hard that you tested and rejected — and explain why with data — is the strongest possible answer. It shows you serve the advertiser rather than the platform, which is exactly what a client is paying for.

The interviewer wants to see a calm, sequenced audit rather than a panic rebuild. Show that you protect what is working, fix measurement first, and only then restructure. Frame your answer as a 30-day plan with a real or realistic example.

• Week 1: measurement and access. Confirm admin access and billing ownership. Audit every conversion action: which are primary, whether counting is set to One or Every, whether duplicates exist (a GA4 import and a Google tag firing for the same event is a classic). Check auto-tagging and the GA4 link.
• Week 1–2: baseline before touching anything. Export 90 days of performance by campaign, device, location and search term so you can prove later what your changes did. Read the change history to learn what the last team tried.
• Week 2: quick, low-risk wins. Add obvious negatives from the search terms report, fix disapproved ads and broken final URLs, correct location settings, and pause keywords that have spent heavily with zero conversions over a long window.
• Week 3–4: structural fixes. Separate brand from non-brand, consolidate thin ad groups so smart bidding gets enough data, and align bid strategies to the business goal.

Example point: “In one inherited account, form submissions were counted twice, so reported CPA was half the real figure. Fixing that first reset the client’s expectations and stopped smart bidding from chasing phantom conversions.”

Note: Close by saying you would share a written audit with a prioritised action list and agreed KPIs by day 30, so the client sees a plan, not just activity.

This question tests integrity as much as technical skill. Broken tracking usually means past reports were wrong, and the interviewer wants to know whether you surface bad news quickly and fix the root cause. Use STAR and be specific about the diagnosis.

• Situation — “Our lead-gen client’s CPA dropped 40% in a week with no change in spend or leads in their CRM.”
• Task — work out whether the improvement was real before anyone celebrated it.
• Action — explain the steps:
  • Compared Google Ads conversions with CRM leads by day and found Google Ads reporting roughly twice as many.
  • Used Tag Assistant and the conversion action diagnostics to see that a website redesign had added a second tag firing on the thank-you page, and that counting was set to Every instead of One.
  • Fixed the tag, set counting to One for leads, and marked the duplicate action as secondary so bidding stopped using it.
  • Told the client the same day, with a corrected CPA trend and an estimate of how long smart bidding would need to recalibrate.
• Result — reported leads matched the CRM within a small margin, bidding stabilised within a couple of weeks, and we added a weekly Ads-versus-CRM reconciliation check.

What to emphasise: you reconciled against a source of truth outside Google Ads, you fixed the process not just the tag, and you communicated before the client found out.

Note: Never present a sudden CPA improvement without first ruling out a tracking change. Saying this unprompted signals real experience.

The interviewer is checking whether you understand diminishing returns and can explain them to a non-specialist without sounding defensive. Your answer should show that you quantified the trade-off and let the client choose with data.

• Situation — “A coaching institute was getting 300 leads a month at ₹450 each and asked for 500 leads at ₹350.”
• Task — set realistic expectations and find the best achievable mix.
• Action — walk through your approach:
  • Checked impression share: most core keywords were already above 80%, so more volume would have to come from broader queries, new locations or new channels, all of which usually cost more per lead.
  • Separated efficiency levers (negatives, landing page speed, better ad copy, lead form assets) from volume levers (broad match, new cities, Demand Gen).
  • Built a simple scenario table showing expected leads and CPL at three budget levels, using marginal rather than average cost.
  • Pushed for lead quality data from their CRM, because a slightly higher CPL with better admission rates was the real goal.
• Result — the client chose the middle scenario: around 420 leads at ₹480, but with admissions up because we optimised towards qualified leads.

Key message: you rarely get both more volume and lower cost from the same account. You can improve efficiency first, then spend that headroom on growth.

Note: Avoid promising both targets to win approval. Interviewers look for candidates who reframe the goal around business outcomes such as cost per admission or profit.

This checks planning discipline and whether you understand how smart bidding reacts to sudden changes. Structure your answer as before, during and after the event, with concrete numbers.

• Before (3–4 weeks out)
  • Pull last year’s festive data: day-by-day conversion rate, CPC and share of monthly revenue.
  • Agree the budget split with the client, for example 40% of the month’s spend in the ten sale days.
  • Make sure the Merchant Center feed shows sale prices and promotions, and approve new ad copy and promotion assets early, because reviews can slow down in peak weeks.
  • Build remarketing and Customer Match lists so you can re-engage past buyers.
• During
  • Raise budgets before the peak, not after campaigns become budget-limited.
  • For a short event where conversion rate is expected to jump, use a seasonality adjustment so smart bidding does not under-bid.
  • Check spend pacing at least twice a day, and watch impression share lost to budget on best sellers.
• After
  • Bring budgets and targets back down promptly so the account does not overspend in a post-sale slump.
  • Report incremental revenue against a pre-sale baseline, not just total sales.

Example point: “We raised budgets 2.5 times for the three peak days, applied a seasonality adjustment, and hit 118% of the revenue target at a ROAS just under plan.”

Note: Mention that you also plan for stock-outs: pausing out-of-stock products quickly stops wasted spend.

The interviewer is testing stakeholder management: can you push back with data while keeping the client on side? A good answer acknowledges the goal behind the request, then reframes it around profit.

Structure your reply in four steps:

1. Understand the motive. Often it is brand visibility, a competitor comparison, or a senior person searching and not seeing the ad. Ask which keywords actually matter to them.
2. Show the cost with their own data. Use absolute top impression rate and a Target Impression Share simulation to show what the top position would cost. For example, “Moving from 45% to 90% absolute top on generic terms roughly doubles CPC, but conversions rise only about 15%.”
3. Offer a targeted compromise. Bid aggressively for top position on brand terms and a handful of high-intent keywords, where it is cheap and protective, and keep conversion-based bidding for everything else.
4. Agree a measure of success. Report cost per acquisition and revenue alongside visibility, so the discussion shifts from position to results.

Example point: “A real estate client wanted the top slot everywhere. We ran a four-week experiment: top-position bidding on 20 priority keywords versus tCPA. The experiment arm cost 60% more per site-visit booking, and the client agreed to keep top-position bidding only on brand.”

Tip: If a founder is worried because they cannot see their own ad, explain that repeated searches without clicking can reduce how often it shows to them, and suggest the Ad Preview and Diagnosis tool instead.

Note: Never simply refuse. Offer an experiment. It turns an opinion into a decision the client owns.

Google Ads runs an auction every time someone searches. You do not simply buy the top position with the highest bid — placement is determined by Ad Rank.

Ad Rank is calculated from:

• Your bid — the maximum you are willing to pay.
• Ad quality — expected click-through rate, ad relevance, and landing page experience, summarised for you as Quality Score.
• Ad Rank thresholds — minimum quality bars an ad must clear to show at all, or in a given position.
• Auction context — the search term, location, device, time, and what else is on the page.
• The expected impact of assets (formerly extensions) such as sitelinks and callouts.

What you actually pay is a second-price-style calculation: the minimum needed to beat the Ad Rank of the competitor below you, divided by your Quality Score — never more than your maximum bid.

The practical consequence, and the point interviewers want: a higher Quality Score lets you pay less for the same position, or hold a better position for the same money. An advertiser with a Quality Score of 8 can outrank one bidding more with a score of 4. That is why relevance work is a cost-reduction lever, not just a hygiene task.

Quality Score is a 1-10 diagnostic at the keyword level, estimating how relevant your ad and landing page are to a search. It is reported as an aggregate of three components, each rated Below Average, Average, or Above Average.

• Expected click-through rate — how likely your ad is to be clicked when shown for this keyword, relative to competitors. The heaviest-weighted component.
• Ad relevance — how closely the ad copy matches the intent of the keyword.
• Landing page experience — relevance of the page content, transparency, ease of navigation, and mobile load speed.

How to improve each:

• Expected CTR — tighter ad groups so the ad speaks directly to the keyword, stronger headlines with the benefit and a clear call to action, and adding assets to increase the ad's footprint.
• Ad relevance — include the keyword in the headline naturally, and split ad groups where one is covering several distinct intents.
• Landing page experience — match the page to the ad promise, improve mobile load time, and remove friction from the form.

Note: Quality Score is a diagnostic, not a KPI. Do not optimise it for its own sake — the auction uses real-time quality signals, and the reported number is a directional summary. Say that, and you will sound like a practitioner.

Match types control how loosely a keyword may match a search query.

• Broad match — running shoes. Matches related searches, synonyms, and queries Google judges relevant, using signals including landing page content and other keywords in the account. Widest reach, least control. Only safe with strong conversion tracking, smart bidding, and an active negative keyword list.
• Phrase match — "running shoes". Matches queries containing the meaning of the phrase, with words before or after. Since the 2021 change it absorbed modified broad match and is now the practical middle ground.
• Exact match — [running shoes]. Matches the same meaning as the term, including close variants such as plurals, misspellings, and reordered words where intent is unchanged. Tightest control, highest intent, lowest volume.

Note: The critical point is that no match type is literal any more — exact match includes close variants, so the search terms report is essential regardless of match type.

A common structure: start with exact and phrase for known converting terms, use broad match with smart bidding to discover new queries, and mine the search terms report continually — promoting winners to exact and adding losers as negatives.

Manual:

• Manual CPC — you set bids yourself. Maximum control, but does not use Google's per-auction signals. Useful for very low-volume accounts where automation cannot learn.

Automated, by objective:

• Maximise Clicks — traffic volume. Reasonable for a brand-new campaign gathering data, but it optimises for clicks, not customers.
• Maximise Conversions — as many conversions as the budget allows, with no cost constraint. Use when volume matters more than efficiency.
• Target CPA — conversions at a specified average cost. The standard choice for lead generation.
• Maximise Conversion Value / Target ROAS — optimises for revenue rather than conversion count. The right choice for e-commerce, where a 500 order and a 50,000 order are not equivalent.
• Target Impression Share — for visibility goals, typically brand defence.

How to choose: smart bidding needs conversion data — roughly 30 conversions in 30 days for Target CPA, and more for Target ROAS. Below that, it cannot learn and you are better off with manual or Maximise Clicks while you build volume.

Note: Two practical points: accurate conversion tracking is a prerequisite, since smart bidding optimises towards whatever you tell it is a conversion; and expect a learning period of one to two weeks after any change, during which you must not keep adjusting.

The hierarchy is account → campaign → ad group → keywords and ads.

• Campaigns control budget, bidding strategy, location, language, network, and schedule. So split campaigns wherever those settings need to differ — by product line, by geography, by margin, or by intent.
• Ad groups hold tightly themed keywords with ads written specifically for them. The rule of thumb is that every keyword in an ad group should be answerable by the same ad.

Why structure matters:

• Relevance and Quality Score. Tight ad groups let the ad match the query closely, which lifts CTR and lowers cost per click.
• Budget control. Budgets are set at campaign level, so lumping high-margin and low-margin products together means you cannot fund them differently.
• Clear reporting. If a campaign mixes intents, the data cannot tell you which is working.

Note: The nuance worth adding is that structure has changed with smart bidding. Very granular single-keyword ad groups used to be best practice; now they can fragment conversion data so thinly that automated bidding cannot learn. The modern approach is consolidating enough to give the algorithm signal while keeping campaigns separate where settings or budget genuinely differ.

Conversion tracking records what happens after the click. Without it you are optimising towards clicks, which is optimising towards spending money.

How it works: you define a conversion action, then place a tag or trigger an event when it occurs — a purchase, a form submission, a call, an app install.

Setting it up correctly:

• Use Google Tag Manager rather than hard-coding tags, so changes do not require a developer and you can preview before publishing.
• Track meaningful actions. A page view of a thank-you page is a weak proxy; a verified lead or a completed purchase is real. Where possible import offline conversions so the algorithm optimises towards qualified leads rather than form fills.
• Set values. A conversion with a value lets you use Target ROAS and lets the algorithm distinguish a large order from a small one.
• Choose count settings deliberately — "every" for purchases, "one" for leads, otherwise a single user filling a form three times inflates results.
• Set a sensible conversion window matched to your real sales cycle.
• Configure enhanced conversions and consent mode to recover measurement lost to cookie restrictions.

Note: Always verify in the real world — submit a test lead and confirm it appears. And beware of double counting when both Google Ads and Analytics conversions are imported.

They differ by where ads appear and, crucially, by intent.

• Search — text ads on Google results pages, triggered by keywords. The user is actively looking, so intent is highest and conversion rates are best. This is where most budgets should start.
• Display — image and responsive ads across millions of partner sites. Users are not searching, so it is a demand-generation and remarketing channel. Cheap clicks, low conversion rates, and easy to waste money on if you do not exclude poor placements.
• Shopping — product listings with image, price, and merchant, driven by a Merchant Center feed rather than keywords. Essential for e-commerce; the feed quality is what determines performance.
• Video — YouTube placements. Strong for awareness and consideration; measurable direct response is harder and needs realistic expectations.
• Performance Max — one campaign serving across every Google surface using automation, driven by asset groups and audience signals rather than keywords.

Note: Performance Max is the one interviewers probe on. The honest assessment is that it can perform well with strong conversion data and good creative assets, but it gives limited visibility into where spend goes and can cannibalise branded search. The practical advice is to add brand negatives, feed it high-quality assets and audience signals, and never run it as your only campaign type.

Negative keywords prevent your ads from showing on specified queries. They are the main tool for stopping wasted spend, and the single highest-return routine task in a search account.

The search terms report shows the actual queries people typed, as opposed to the keywords you bid on. The two are never identical, and the gap is where money leaks.

The routine:

• Review it regularly — weekly for a high-spend account, monthly for a small one.
• Sort by cost and look for queries with spend and no conversions. Add irrelevant ones as negatives.
• Look for winners too. A converting query not in your account should be promoted to its own keyword, where you can control its bid and write specific copy.
• Add negatives at the right level. Campaign or ad group level for targeted exclusions; a shared negative list applied across campaigns for universal ones such as "free", "jobs", "salary", and competitor names where you do not want to bid.

Common negative themes: job seekers, students researching, DIY intent, free intent, and — often overlooked — a cross-negative structure that stops your broad campaign stealing traffic from your exact campaign.

Note: Negative match types work the same way but there is one trap: negative phrase and exact do not include close variants, so misspellings must be added separately.

Separate the metrics that describe the business from those that only diagnose the account — conflating the two is the most common mistake.

Business metrics — what you are judged on:

• Conversions and conversion rate — is it producing outcomes?
• Cost per acquisition (CPA) — the key metric for lead generation.
• Return on ad spend (ROAS) — the key metric for e-commerce.
• Total conversion value and profit, where you can get it. ROAS on revenue can hide a loss if margins differ by product.

Diagnostic metrics — useful for finding the cause, not for reporting success:

• Impressions and impression share — reach, and how much you are missing to budget or to rank.
• CTR — ad relevance and copy strength.
• Average CPC — competitiveness and Quality Score effects.
• Search terms and Quality Score — relevance hygiene.

Note: The answer interviewers are hoping for is that CTR and impressions are not goals. A campaign can have an excellent CTR and lose money. Equally, be ready to say that a rising CPA is not automatically bad — if volume grew and the CPA is still under target, that is usually the right trade. Knowing which direction to push depends on whether the business is constrained by efficiency or by volume.

Ad assets, previously called extensions, add extra information beneath your ad. They are free to add, they increase the ad's size on the page, and expected asset impact is a factor in Ad Rank — so they improve position and CTR at no additional cost per click.

• Sitelinks — extra links to specific pages, letting one ad offer several entry points.
• Callouts — short non-clickable benefits such as "Free Shipping" or "24/7 Support".
• Structured snippets — a header and list, such as Brands or Courses.
• Call assets — a phone number, valuable on mobile and trackable as a conversion.
• Location, price, promotion, image, and lead form assets for their respective cases.

A Responsive Search Ad (RSA) is now the only search ad type. You supply up to 15 headlines and 4 descriptions, and Google assembles and tests combinations per auction.

How to write them well: make headlines genuinely distinct rather than fifteen rewordings of one idea, cover different angles — benefit, offer, credibility, call to action — and use pinning sparingly. Pinning gives control but sharply reduces the combinations available, which undercuts the point.

Note: Ad Strength is a guideline, not a ranking factor. Do not damage good copy to turn a rating from Good to Excellent — a frequent trap for people new to the platform.

Location targeting in Google Ads has two options that decide who counts as being in your target area:

• Presence — shows ads to people who are in, or regularly in, your targeted locations.
• Presence or interest — also includes people who have shown interest in the location, for example someone in Delhi searching “hotels in Goa”. This is the default in many campaign types.

When it matters:

• Local service businesses such as a clinic in Pune, a home-cleaning service or a coaching centre usually want Presence. With the default, a clinic can pay for clicks from people in other cities who will never visit.
• Travel and hospitality often want Presence or interest, because the buyer is outside the destination.
• Exclusions should normally use Presence too, so you exclude people physically in an area you cannot serve.

How to check whether the setting is costing money: open the location report and compare the “user location” view with targeted locations. If a noticeable share of spend comes from outside your service area, switch to Presence.

Related settings to review at the same time:

• Radius targeting around a store, and location bid adjustments or separate campaigns for high-value cities if performance varies a lot.
• Language targeting — Google uses signals such as query language and user settings. For India, adding Hindi and other regional languages can help if your ads and landing pages support them.

Note: Checking location options is one of the fastest wins in an account audit. Mention it and interviewers know you have done real audits.

Smart bidding is the default choice for most mature accounts, but there are specific situations where click-based strategies are the better tool.

Manual CPC — you set maximum bids per keyword or ad group.

• No reliable conversion data yet — a new account, or tracking still being fixed. Conversion-based bidding has nothing to learn from.
• Very low volume niches — a B2B industrial product with a handful of conversions a month may never give smart bidding enough signal.
• Tight control during tests — for example, isolating the effect of new ad copy without the bid strategy also shifting.
• Brand campaigns where you want to cap CPC strictly.

Maximize Clicks — Google sets bids to get as many clicks as possible within budget.

• Useful to gather traffic and data quickly for a new campaign, ideally with a maximum CPC limit so a few expensive clicks do not eat the budget.
• Useful for traffic-driven goals such as content or awareness, where a click is the outcome.
• The risk: it chases cheap clicks, not valuable ones, so traffic quality can fall.

A typical progression:

1. Launch on Manual CPC or Maximize Clicks with a bid cap while tracking is validated.
2. Once the campaign has a steady flow of conversions, move to Maximize Conversions.
3. After a stable CPA emerges, add a target CPA close to the actual recent CPA.

Note: Enhanced CPC, the old middle ground, has been phased out on Search and Display, so do not propose it as your main strategy. Say you would move straight from manual bidding to Maximize Conversions.

Maximize Conversions spends your budget to get as many conversions as possible. Adding a target CPA tells it to aim for an average cost per conversion instead. The transition needs care, because a bad target either starves the campaign or wastes money.

When to add a target:

• The campaign has run on Maximize Conversions for a few weeks and is past its learning period.
• It records a steady flow of conversions. Google does not enforce a hard minimum, but more data gives more stable results; around 30 or more conversions a month is a common rule of thumb.
• Conversion tracking has been checked and conversion delays are understood.

How to choose the target:

1. Start from the actual CPA of the last 2–4 weeks, not the CPA you wish you had.
2. Set the first target at or slightly above that figure. If recent CPA is ₹600, start around ₹600–650.
3. Tighten in small steps of about 10–15% every couple of weeks, checking that volume holds.
4. Keep your break-even CPA in mind as a ceiling. It comes from unit economics, not from Google Ads.

What goes wrong:

• Target far below actual CPA — impressions and spend collapse because the system bids too low to win auctions.
• Changing targets every few days — the system never settles.
• Budget too small for the target — if the daily budget covers only one or two conversions, the strategy has little room to optimise.

Note: Use the bid strategy report and target CPA simulator to check how volume changes at different targets before committing.

Target ROAS (return on ad spend) is an optional target within Maximize Conversion Value. It tells Google to aim for a certain amount of conversion value per rupee spent, expressed as a percentage.

Worked example: a target ROAS of 400% means you want ₹4 of conversion value for every ₹1 spent. At ₹50,000 spend, the system aims for about ₹2,00,000 in reported value.

Conversion values are the foundation. The strategy can only be as good as the values it optimises:

• E-commerce — pass the actual order value dynamically through the tag or the GA4 purchase event. If margins vary widely by category, consider passing profit or using conversion value rules, so a ₹1,000 order of a 60% margin item is not treated the same as a 10% margin one.
• Lead generation — assign values that reflect lead quality. For example, a demo request at ₹2,000 and a newsletter sign-up at ₹50, or import CRM stages with different values.
• Currency and tax — be consistent about whether values include GST and shipping.

Setting the target:

1. Look at actual ROAS over the last few weeks, allowing for conversion delay.
2. Start near that figure, then raise it gradually.
3. Remember that a higher target ROAS means lower volume. The system becomes more selective.

Common mistakes: launching tROAS with no value history, setting the target from a spreadsheet instead of actual performance, and forgetting that returns and cancellations inflate reported value unless you adjust for them.

Note: Frame ROAS targets against profit. A 400% ROAS is excellent at 40% margin and loss-making at 20%.

Targets in Google Ads should come from the business model, not from industry benchmarks. Interviewers like candidates who can do this arithmetic live.

Break-even ROAS for e-commerce

• Formula: break-even ROAS = 1 / gross margin (margin after product cost, shipping and payment fees).
• Example: margin is 25%. Break-even ROAS = 1 / 0.25 = 4, or 400%. Below 400%, each sale loses money on the first order.
• If you need a 10% contribution margin after ads, ad spend can be only 15% of revenue, so the target is 1 / 0.15 ≈ 667%.

Break-even CPA for e-commerce

• Average order value ₹2,000 × 40% margin = ₹800 gross profit per order. Break-even CPA is ₹800.
• If customers reorder, you can justify a higher CPA using lifetime value. With an average of 2.5 orders, gross profit per customer is ₹2,000, so a first-order CPA of ₹1,000 may be acceptable.

Break-even cost per lead for lead generation

• Deal value ₹50,000 at 30% margin = ₹15,000 profit per sale.
• If 10% of leads become customers, each lead is worth ₹1,500. Break-even CPL is ₹1,500.
• A realistic target sits below that, for example ₹1,000–1,200, to leave room for profit and sales costs.

Note: Mention whether you are using first-order or lifetime value. Mixing them up is the most common reason ad targets and finance forecasts disagree.

Both let you manage several campaigns together, but they solve different problems.

Portfolio bid strategy — one automated strategy (for example target CPA or target ROAS) shared across multiple campaigns.

• Pooled data: low-volume campaigns benefit from the conversion history of the whole group, which helps stability.
• One target, flexible distribution: the portfolio aims for the target on average. One campaign may run at ₹700 CPA and another at ₹400 if the blend meets the target, which lets the system shift effort to where conversions are cheaper.
• Bid limits: portfolio strategies let you set maximum and minimum CPC bid limits for target CPA and target ROAS, which standard campaign-level strategies do not. Use limits carefully, because they can restrict performance.
• Use when: several campaigns share the same goal and economics, such as regional campaigns selling the same product.
• Avoid when: campaigns have different value per conversion, such as brand and non-brand, or a high-margin and a low-margin product line.

Shared budget — one daily budget used by several campaigns.

• Google shifts spend towards campaigns with more opportunity on a given day.
• Use when: you have many small campaigns with the same priority and do not want to micromanage their budgets.
• Avoid when: one campaign must get a guaranteed share. A broad generic campaign can absorb the whole shared budget and starve brand.

Note: A good rule: group campaigns only when they share the same goal and the same value of a conversion. Separate budgets for anything you need to control or report on independently.

Every conversion action in Google Ads is either primary or secondary, and that choice decides whether smart bidding optimises towards it.

• Primary actions appear in the Conversions column and are used by smart bidding (for goals the campaign is using).
• Secondary actions appear only in the All conversions column. They are recorded for observation but do not steer bids.

Conversion goals group actions by category, such as Purchases, Leads or Phone calls. Goals can be set as account default, which every campaign uses unless told otherwise, or campaign-specific, where one campaign optimises to a different goal.

Worked example: a car dealer tracks test-drive bookings, brochure downloads, calls and page scrolls.

• Primary: test-drive bookings and qualified calls, the actions that lead to sales.
• Secondary: brochure downloads and scroll depth, which are useful to watch but would pull bidding towards cheap, low-intent users if they were primary.

Common mistakes:

• Leaving micro-conversions such as page views as primary, which inflates conversion counts and makes CPA look artificially low.
• Having two primary actions for the same event (for example a Google tag conversion and a GA4 import of the same purchase), which double counts.
• Changing primary actions without expecting a learning period. It changes what the algorithm is optimising, so performance can wobble for a while.

Note: A strong answer ties this back to business outcomes: primary actions should be the ones closest to revenue that still have enough volume for bidding to learn from.

Enhanced conversions improve conversion measurement by sending hashed first-party customer data, such as email address, phone number, or name and address, along with a conversion. Google matches this to signed-in Google accounts to attribute conversions that cookies alone would miss.

How it works:

1. A user converts on your site, for example by submitting a checkout or lead form.
2. The tag captures the customer data and hashes it with SHA-256 before it is sent, so raw personal data is not transmitted in plain text.
3. Google matches the hashed data with its logged-in users and credits the conversion to the ad interaction.

Two flavours:

• Enhanced conversions for web — improves attribution of online conversions such as purchases.
• Enhanced conversions for leads — for businesses that close deals offline. The hashed email from the form is later matched to an offline conversion upload, often as an alternative to storing the GCLID.

Why it matters:

• Browser privacy restrictions, cookie consent and cross-device journeys all lose conversions. Enhanced conversions recover part of that gap.
• More complete conversion data means smart bidding learns from a truer signal.

Setup options: the Google tag with automatic or code-based detection, Google Tag Manager, or the Google Ads API. Check the diagnostics tab afterwards for match rate and errors.

Compliance: you must follow Google’s customer data policies, update your privacy notice, and respect consent requirements, which in India means keeping the Digital Personal Data Protection Act in mind.

Note: Interviewers like to hear that enhanced conversions improve measurement, not targeting. It does not create an audience list, which is what Customer Match does.

For lead-gen businesses such as real estate, education, insurance or B2B software, a form fill is not the sale. If Google Ads only sees form fills, smart bidding optimises for cheap leads, including junk ones. Offline conversion import sends the downstream outcome, such as a qualified lead or a closed sale, back to Google Ads.

How it works:

1. Capture the click identifier. With auto-tagging on, each ad click carries a GCLID in the URL. Store it in a hidden form field and pass it to the CRM with the lead. Alternatively, use enhanced conversions for leads, which matches on hashed email or phone.
2. Define offline stages. Create conversion actions such as “Qualified lead”, “Site visit done” and “Booking”, with values if possible.
3. Upload outcomes. Use a scheduled file upload, a CRM integration (for example Salesforce, HubSpot or Zoho through connectors) or the API. Uploads must happen within the conversion window, which can be up to 90 days after the click.
4. Switch optimisation. Once volume is steady, make the qualified stage primary and move the raw form fill to secondary.

Worked example: an ed-tech company paying ₹300 per lead finds only 5% of leads enrol, so the real cost per enrolment is ₹6,000. After importing “Counselling attended”, the cost per lead rises to ₹380 but enrolment rate doubles to 10%, bringing cost per enrolment down to ₹3,800.

Practical tips:

• Pick the deepest stage that still gives enough monthly volume for bidding.
• Upload frequently, ideally daily, so bidding reacts quickly.
• Reconcile imported counts against the CRM every week.

Note: This is one of the highest-impact improvements in lead-gen accounts. Mentioning it shows you optimise for revenue, not just lead volume.

Attribution decides how credit for a conversion is shared across the ad interactions that led to it. Conversion windows decide how long after an interaction a conversion can still be counted.

Attribution models available today:

• Data-driven attribution (DDA) — the default for most conversion actions. It uses your account’s path data to estimate how much each interaction, such as a generic search click followed by a brand click, contributed.
• Last click — gives all credit to the final ad click. Simple, but it over-credits brand and remarketing and under-credits upper-funnel campaigns.
• Older rule-based models such as first click, linear, time decay and position-based have been retired in Google Ads.

Why it affects bidding: smart bidding optimises towards the credited conversions. Under last click, a generic keyword that starts many journeys looks weak and gets bid down. Under DDA, it receives fractional credit, so bidding values it more fairly.

Conversion windows:

• Click-through windows can be set from 1 to 90 days, with 30 days a common default.
• Match the window to the real sales cycle. A food delivery order happens within minutes; a home loan may take weeks.
• Engaged-view and view-through windows apply to video and display interactions and should be kept short and reported separately.

Worked example: a user clicks “best laptop under 60000”, returns three days later via a brand search and buys. Last click gives 100% to brand. DDA might give roughly half to the generic click, so the generic campaign’s CPA looks better, and bidding can invest in it.

Note: Changing attribution models changes historical-looking numbers and can shift bidding. Tell the client before you switch, and compare before and after on the same basis.

Clients often see different conversion numbers in Google Analytics 4 and Google Ads for the same period. Both can be correct: they are measuring differently.

Main reasons for the gap:

• Attribution scope. Google Ads only looks at Google ad interactions. GA4 attributes across all channels, so a conversion credited to a paid click in Google Ads may be credited to email or organic search in GA4.
• Date of conversion. Google Ads reports conversions against the date of the ad click by default. GA4 reports them on the date they happened. A click on 30 June and a purchase on 2 July land in different months.
• Counting method. Google Ads may count One per click for leads while GA4 counts every event, or the reverse.
• Conversion windows and models. Different lookback windows and attribution settings change totals.
• View-through and cross-device conversions are included in some Google Ads columns but not in standard GA4 reports.
• Consent and modelling. Each platform models missing data differently.
• Import delay. GA4 conversions imported into Google Ads can take hours to appear.

How to handle it:

1. Choose one source for bidding. Many advertisers use the Google Ads tag with enhanced conversions for bidding because it is faster and click-based, and use GA4 for cross-channel analysis.
2. Never have the same event as primary from both sources, or you double count.
3. Agree which report is the official number for client reporting, and reconcile both against a backend source such as orders or CRM leads.
4. Accept a stable gap; investigate a sudden change in the gap.

Note: The backend system, not either Google tool, is the real source of truth. Saying this shows maturity.

Because Responsive Search Ads (RSAs) assemble combinations on the fly, the old method of running two static ads and comparing CTR no longer maps cleanly. You need a mix of asset-level insight and controlled experiments.

1. Read asset-level data

• The assets report shows impressions and, where available, performance data for each headline and description, plus combination reports showing which assets served together.
• Use it to spot weak assets that rarely serve, and replace them with genuinely different messages, not near-copies.

2. Test themes, not single words

• Group headlines by angle: price (“Plans from ₹499”), trust (“4.6 rated by 1 lakh users”), speed (“Delivered in 10 minutes”) or offer (“Flat 20% off today”).
• Swap a whole theme in or out and measure the effect on conversion rate, not only CTR.

3. Use ad variations or campaign experiments

• Ad variations let you apply a text change, such as replacing “Free Trial” with “Free Demo”, across many ads and split traffic.
• Custom experiments split traffic between the original and a draft with different RSAs, giving a cleaner comparison.

4. Use pinning sparingly

• Pin only for legal disclaimers, brand names or regulated claims. Every pin reduces the combinations Google can test.

What to measure: conversions per impression is a better single metric than CTR, because it captures both the click and the landing experience.

Note: Ad Strength is guidance for asset variety, not a ranking factor. Do not chase “Excellent” at the expense of a message that actually converts.

When you add audience segments to a campaign or ad group, you choose how they are applied.

• Targeting — restricts reach. Ads show only to people in the selected audiences (subject to other targeting such as keywords or placements).
• Observation — does not restrict reach. Ads show to everyone your other targeting allows, but you get reporting for those audiences and can apply bid adjustments on Manual CPC.

When to use Observation:

• On Search campaigns in most cases. Keywords already capture intent, and you want to learn which audiences, such as past site visitors or in-market for home loans, convert better without losing volume.
• When a campaign is new and you are not yet sure which segments matter.
• With smart bidding, observation audiences also give you visibility, while the algorithm uses audience signals in its own bidding anyway.

When to use Targeting:

• Display and Video campaigns, where audiences are often the main targeting method.
• Remarketing Search campaigns (RLSA) where you deliberately bid on broader keywords only for past visitors, such as “laptops” only for people who viewed your laptop pages.
• When budget is limited and you must focus on the highest-value segment.

Worked example: an insurance brand adds “In-market: Life insurance” as Observation on Search. After a month it converts at twice the account average, so they create an RLSA-style campaign targeting that segment on broader generic keywords.

Note: Also mention optimised targeting on Display and Demand Gen, which can expand beyond your chosen audiences. Turn it off when you need strict audience control.

Both let you treat people who already know you differently from strangers, which is usually where the cheapest conversions are.

Remarketing lists for search ads (RLSA)

• Built from website or app visitors via the Google tag or GA4 audiences, for example “Viewed product, did not buy, last 30 days”.
• Search and YouTube need larger lists than Display before they can serve, so build lists early.
• Uses:
  • Bid higher or keep a higher target for past visitors on your core keywords.
  • Bid on broader, otherwise unprofitable terms only for past visitors.
  • Show different copy, such as “Complete your booking today” or an exclusive offer.

Customer Match

• You upload first-party customer data, such as emails and phone numbers, which is hashed and matched to Google accounts.
• Works across Search, Shopping, YouTube, Gmail, Display and Performance Max as a signal or segment.
• Eligibility depends on account history and policy compliance, and you must have consent to use the data.
• Uses:
  • Exclude existing customers from acquisition campaigns.
  • Upsell or cross-sell, for example a bank targeting existing savings customers with credit card ads.
  • Seed audience signals in Performance Max and lookalike segments in Demand Gen.

Worked example: an Indian D2C skincare brand uploads its 80,000-customer list. It excludes them from the “new customer” campaign, runs a replenishment campaign for customers who bought 60 days ago, and uses high-value buyers as a Performance Max signal.

Note: Refresh Customer Match lists regularly. A list uploaded once and forgotten slowly goes stale as customers change emails and phone numbers.

Shopping and Performance Max for retail use the Merchant Center product feed instead of keywords. Google reads your titles and attributes to decide which searches your products match, so the feed is effectively your keyword list and ad copy.

Core attributes most products need:

• id, title, description, link, image_link
• price and availability, which must match the landing page
• brand and gtin for products with a manufacturer-assigned barcode, or mpn where appropriate
• condition when products are used or refurbished
• Variant attributes such as size, color and item_group_id for apparel

Feed optimisation levers:

1. Titles — put the most searched attributes first. “Nike Men’s Revolution 6 Running Shoes, Black, Size 9” beats “Revolution 6”.
2. Product types and categories — accurate categories help matching and reporting.
3. Images — clean, high-quality, no promotional overlays.
4. Custom labels (0–4) — tag margin band, best sellers, seasonality or clearance, so you can split campaigns or asset groups by business value.
5. Sale price and promotions — show discounts in the ad during sale events.

Diagnostics: check the Merchant Center diagnostics regularly for disapprovals, such as price mismatch, missing GTIN or policy issues, and fix them quickly, because disapproved products simply stop serving.

Note: Use supplemental feeds or feed rules to improve titles without editing the website’s product catalogue. Interviewers value knowing this practical workaround.

Even with Performance Max available, many advertisers still run Standard Shopping for more control, reporting and negative keywords. Structure is how you get that control.

Product groups

• Subdivide inventory by category, brand, item ID, product type or custom label.
• Bid differently on each group. On Manual CPC you might bid ₹12 on best sellers and ₹5 on long-tail items.
• Always keep an “everything else” group so new products are not left out.

Segmentation by business value

• Use custom labels to split products by margin band or performance tier, for example high-margin heroes, average, and low-margin items.
• Run separate campaigns or product groups with different targets, such as target ROAS 300% for high margin and 600% for low margin.

Campaign priority (low, medium, high)

• When the same product is in several Shopping campaigns, the higher priority campaign bids first, regardless of bid.
• Classic query-sculpting setup:
  • High priority, low bids — catches generic queries such as “running shoes”, with negatives for brand and model terms.
  • Medium priority, mid bids — catches brand queries such as “nike running shoes”, with negatives for specific models.
  • Low priority, high bids — catches specific, high-intent queries such as “nike revolution 6 size 9”.
• This lets you pay more for specific queries without writing keywords.

Negatives and search terms — review Shopping search terms regularly and add negatives for irrelevant queries, such as “free”, “second hand” or competitor terms you do not want.

Note: Campaign priority only works between Standard Shopping campaigns. When Performance Max covers the same products, Google decides which campaign serves using its own rules, currently mainly Ad Rank, so plan product overlap deliberately.

Performance Max (PMax) runs across Google’s inventory from one campaign. You influence it through inputs rather than keyword-level controls, so knowing each input is essential.

• Asset groups — collections of headlines, descriptions, images, videos and logos, plus a set of listing groups for retail. Build one per theme, such as product category or audience, so creative matches intent. Aim for enough variety to fill every format, including video, or Google may auto-generate some.
• Audience signals — hints such as Customer Match lists, website visitors, custom segments based on search terms, and interests. They guide where the system starts looking but do not restrict reach. Your own first-party data is usually the strongest signal.
• Search themes — phrases you supply to tell PMax which search topics are relevant, useful when a new campaign lacks data. They are signals, not keywords, and they do not override exact match keywords in Search campaigns.
• Brand exclusions — stop PMax from serving on your own brand queries, so brand traffic stays in a Search campaign where you can see and control it.
• Negative keywords and URL controls — use negatives where the account supports them, and final URL expansion settings to decide whether Google can send traffic to other pages on your site.

Reporting to use: asset group and asset performance, search term insights, channel-level views, and audience insights. Reporting has improved but is still less granular than Search.

Worked example: a furniture retailer splits PMax into Sofas, Beds and Décor asset groups, adds a past-buyer Customer Match signal, excludes its brand name, and uses a “high margin” custom label to run a separate PMax campaign with a lower ROAS target.

Note: The biggest PMax lever is the quality of your conversion data and feed. Inputs matter more than tweaks.

Demand Gen campaigns are designed to create interest and drive action on Google’s visual, feed-based surfaces. They replaced the older Discovery campaigns.

Where they run: YouTube (including in-stream, in-feed and Shorts), the Discover feed and Gmail, with Google extending the campaign type to more inventory over time. You can choose channels or let Google optimise across them.

What makes them different:

• Formats: image, carousel and video ads in one campaign, plus product feeds for retailers.
• Audiences: strong audience controls, including lookalike segments built from your first-party lists, which Display campaigns do not offer in the same way.
• Bidding: conversion-focused strategies such as Maximize Conversions, target CPA, or value-based bidding, as well as click-based options.
• Creative testing: ad group-level control makes it easier to compare creatives and audiences than in Performance Max.

When to choose Demand Gen over Display:

• You have strong visual and video creative, for example fashion, beauty, food or consumer apps.
• You want mid-funnel growth: reaching people who look like your customers before they search.
• You care about placement quality. YouTube and Discover are generally higher quality environments than the long tail of Display sites and apps.

When Display still fits: broad, low-cost reach, contextual targeting by topic or placement, and simple remarketing at scale.

Worked example: an Indian fashion brand uses Demand Gen with a lookalike of its top 10% customers, short vertical videos for Shorts, and a product feed. It measures success on new-customer purchases, not on clicks.

Note: Expect higher CPA than Search. Demand Gen creates demand; judge it on incremental impact and assisted conversions too.

Video campaigns run on YouTube and Google video partners. The right format and bid strategy depend on whether the goal is awareness, consideration or action.

Main formats:

• Skippable in-stream — plays before, during or after a video; viewers can skip after 5 seconds. Flexible and the most widely used.
• Non-skippable in-stream — short ads viewers must watch, typically up to 15 seconds. Good for a complete message.
• Bumper ads — non-skippable, up to 6 seconds. Cheap reach and frequency for simple messages.
• In-feed video — appears in YouTube search results and the home feed; the user clicks to watch.
• Shorts ads — vertical video between Shorts; creative should be made for vertical, not cropped from landscape.

Practical points:

• Conversion-focused video buying has been moving into Demand Gen, so check the current campaign options when you plan action goals.
• Brand the first five seconds; most viewers decide to skip then.
• Use frequency caps and content suitability settings to avoid overexposure and unsafe placements.
• Measure with Brand Lift studies where budget allows, because clicks under-represent video’s effect.

Worked example: a new UPI app launch runs bumpers plus non-skippable ads on Target CPM for reach in the first two weeks, then skippable in-stream on a conversion-based strategy for installs.

Note: Do not judge video on last-click CPA alone. Look at view-through conversions, search lift for the brand, and Brand Lift results.

App campaigns promote Android and iOS apps across Google Search, Google Play, YouTube, Discover and the Display Network from a single campaign. There are no keywords or individual ads: you supply assets and goals, and Google assembles ads.

Campaign subtypes:

• App installs — acquire new users.
• App engagement — bring existing users back to take an action.
• App pre-registration — for Android apps before launch.

Inputs you control: text lines, images, videos (landscape, square and vertical), HTML5 assets, location and language, budget, and bidding.

Optimising beyond installs:

1. Track in-app events through Firebase or a mobile measurement partner such as AppsFlyer or Adjust, and import events like registration, KYC completed or first purchase.
2. Move up the funnel of goals: start with target cost per install to build volume, then switch to target CPA on a meaningful in-app action, then to target ROAS once you have enough purchase value data.
3. Choose events with volume. An event that happens only a few times a day will not give the system enough signal; pick the deepest event that still has steady volume.
4. Refresh creative regularly and use the asset report to replace low performers, especially videos.

Worked example: a fintech app pays ₹40 per install, but only 10% complete KYC, so cost per KYC is ₹400. Switching to target CPA on KYC completion at ₹350 raises cost per install to ₹55, but KYC rate rises to 18%, bringing cost per KYC down to about ₹306.

Note: Mention iOS measurement limits caused by Apple’s privacy framework, and the need to validate results against your MMP data rather than relying only on Google Ads reports.

The Display Network reaches millions of sites and apps, which means reach is easy and waste is also easy. Good Display management is mostly about targeting and exclusions.

Targeting options:

• Audience segments — remarketing lists, Customer Match, custom segments (built from keywords, URLs or apps), in-market, affinity, life events and detailed demographics.
• Content targeting — keywords (contextual), topics, and specific placements such as named websites or YouTube channels.
• Optimised targeting — lets Google expand beyond your chosen segments to find likely converters. Good for scale, but turn it off when you need strict control.
• Demographics — age, gender, parental status and household income where available.

Brand safety and quality controls:

• Content suitability — choose the inventory type and exclude sensitive content categories such as tragedy or sensational content.
• Placement exclusions — maintain an account-level exclusion list of poor sites, apps and channels found in the placements report.
• App category exclusions — useful because accidental clicks in games and some utility apps waste budget.
• Frequency management and a sensible audience mix to avoid bombarding the same users.

Optimisation routine: review the “where ads showed” placement report weekly, exclude placements with high spend and zero conversions or suspicious CTR, and check conversion rate by placement type.

Worked example: a bank running Display for credit cards found 30% of spend on mobile games with a very high CTR but almost no applications. Excluding those app categories cut CPA by over 25%.

Note: Very high CTR on Display is usually a warning sign of accidental clicks, not a success.

In most campaign types, the budget you enter is an average daily budget, not a hard daily limit.

• On any single day, Google can spend up to two times the average daily budget to take advantage of high-traffic days.
• Over a month, you will not be charged more than about 30.4 times the average daily budget (the average number of days in a month). If Google overdelivers beyond that, you are not charged for the excess.

Why this matters: a client with a strict ₹3,00,000 monthly cap should not be surprised by a ₹20,000 day on a ₹10,000 daily budget. Explain the averaging upfront.

Pacing a monthly budget:

1. Set the starting daily budget as monthly budget ÷ days in the month. ₹3,00,000 ÷ 30 = ₹10,000.
2. Check pacing regularly: compare actual spend to date with expected spend to date. By day 15 you should be near ₹1,50,000.
3. Recalculate: remaining budget ÷ remaining days. If you spent ₹1,80,000 by day 15, the new daily budget is ₹1,20,000 ÷ 15 = ₹8,000.
4. Weight by performance: shift budget to days, campaigns and seasons that convert best, not just evenly.

Tools that help:

• Budget reports and forecasts in the interface.
• Automated rules or scripts to alert or pause when spend crosses a threshold.
• Shared budgets for low-priority campaigns.

Watch-outs: big mid-month budget changes can trigger a new learning phase for smart bidding, and a campaign that is “limited by budget” on smart bidding may be leaving cheap conversions on the table.

Note: Mention that mid-month budget edits change the monthly limit calculation, so large changes should be planned rather than made daily.

Search impression share is the percentage of eligible impressions your ads actually received. The impressions you missed are split into two diagnostic metrics:

• Search lost IS (budget) — impressions missed because the budget ran out or Google throttled delivery to pace your budget.
• Search lost IS (rank) — impressions missed because your Ad Rank was too low, due to bid, ad quality or thresholds.

How to read them together:

Worked example: a brand campaign shows 92% impression share with 6% lost to budget. That is cheap, protective traffic being missed, so the budget should rise. A generic campaign shows 35% impression share with 55% lost to rank. Raising budget will not help; better relevance or higher bids would.

Important caveats:

• With smart bidding, some “lost to rank” is deliberate: the system chooses not to compete in auctions that would miss your target.
• 100% impression share is rarely the right goal on generic terms; the last few percent are usually the most expensive.
• Check top and absolute top impression share as well, because being present but always in lower positions can limit clicks.

Note: Always pair impression share with CPA or ROAS. Extra impressions are only worth buying if the incremental conversions are profitable.

With broad and phrase match, one query can be eligible to match many keywords across ad groups and campaigns. Knowing which one wins matters for reporting, landing pages and cost.

How Google prioritises (in simplified form):

1. Identical exact match wins. If the query is identical to an exact match keyword, including close spelling variants, that keyword is preferred, as long as it is eligible to serve.
2. Otherwise, relevance. Google considers which eligible keyword is most relevant, using the keyword’s meaning, the other keywords in the ad group and the landing page.
3. Then Ad Rank. Among similarly relevant keywords, the one with the highest Ad Rank serves.

Paused keywords, keywords blocked by negatives, and those limited by budget or targeting are not eligible.

Why it matters: if “cheap flights to Goa” triggers your generic “flights” ad group, the user lands on a generic page and conversion rate falls. Budget also gets split unpredictably.

How to control query routing:

• Add the important query as exact match in the ad group with the most relevant ad and landing page.
• Cross-negatives: add specific terms as negatives in broader ad groups so traffic flows to the specific one. For example, add “goa” as a negative in the generic flights ad group.
• Separate brand and non-brand campaigns, with brand terms as negatives in non-brand campaigns.
• Use brand exclusions and negatives in Performance Max so it does not absorb queries your Search campaigns should handle.
• Check the search terms report by ad group to spot the same query appearing in multiple places.

Note: Do not over-sculpt with dozens of cross-negatives; with smart bidding, a simpler structure with enough data per campaign often performs better.

Brand campaigns bid on your own company or product names. Non-brand campaigns bid on generic or category terms such as “term insurance plans”. They behave so differently that mixing them hides the truth.

Why separate them:

• Very different economics. Brand clicks might cost ₹5 with a 15% conversion rate; generic clicks might cost ₹60 at 3%. A blended CPA makes generic look better than it is.
• Budget control. Brand should never be limited by budget; generic can absorb unlimited spend. In one campaign, generic can starve brand.
• Different targets. Brand might run on Target Impression Share for protection; generic on target CPA or ROAS.
• Cleaner reporting for how much new demand you are actually creating.

Is brand search incremental? Many brand clicks would have come through organic results anyway. Arguments for running it:

• Competitors bidding on your name can take the top slot above your organic listing.
• You control the message, offers and sitelinks.
• Brand CPCs are usually cheap because of high Quality Score.

How to test it:

1. Check auction insights: if no competitors appear on your brand terms, the risk of pausing is lower.
2. Run a geo holdout: pause brand ads in a set of comparable cities for a few weeks and compare total brand-driven conversions (paid plus organic) against control cities.
3. Measure the combined result, not just the paid channel. If organic picks up nearly all the lost clicks, the paid brand spend was largely non-incremental.

Note: Report non-brand performance on its own to leadership. It is the clearest measure of how well paid search is growing the business.

Campaign experiments let you test a change against the original campaign on a split of traffic, instead of changing everything and comparing before-and-after, which is distorted by seasonality.

Designing the experiment:

1. One clear hypothesis. “Switching from target CPA to Maximize Conversion Value with values imported from the CRM will increase revenue at a similar cost.”
2. Change one thing. If you change the bid strategy and landing page together, you cannot tell which caused the result.
3. Split traffic evenly, usually 50/50, using a search-based or cookie-based split. Cookie-based keeps a user in one arm, which is better for tests where users return several times.
4. Pick the success metric in advance, such as conversions, CPA, or conversion value, plus a guardrail such as cost.
5. Decide the duration upfront. Allow for the learning period and at least a couple of full conversion cycles; for most accounts that means four to six weeks.

Reading the result:

• Use the experiment report’s confidence indicators. A difference that is not statistically significant is not a win.
• Ignore the first days while bidding learns.
• Check that both arms had similar budgets and were not budget-limited differently.
• Look at conversion lag: in lead-gen, late conversions can change the outcome.

Common mistakes: stopping as soon as one arm looks ahead, editing the original campaign during the test, running too little traffic to ever reach significance, and applying a “winner” without checking business metrics outside Google Ads.

Note: Also mention other test types: ad variations for copy, Performance Max experiments, and video experiments. Then explain that you choose the tool that isolates your hypothesis.

Interviewers ask this to see whether you work systematically or just react. A strong answer separates monitoring from optimisation, and explains that smart bidding needs time, so not everything should be changed daily.

Daily (10–20 minutes per account)

• Spend pacing against budget, and any sudden spike or drop in spend, clicks or conversions.
• Disapproved ads, Merchant Center product disapprovals and billing issues.
• Tracking health: are conversions still recording? A flat zero usually means a broken tag.

Weekly

• Search terms report: add negatives and promote strong queries to keywords.
• Performance by campaign against CPA or ROAS targets; adjust targets or budgets in small steps.
• Impression share lost to budget and rank on key campaigns.
• Placements on Display and video; exclude poor ones.
• Asset performance: replace weak RSA assets and creative.

Monthly

• Device, location, audience and time-of-day analysis to spot structural opportunities.
• Landing page and conversion rate review with the web or product team.
• Auction insights to track competitor movement.
• Plan and review experiments; reconcile conversions with CRM or backend.
• Review recommendations and Optimization Score, applying only those that fit the goal.

Quarterly

• Account structure review, budget reallocation across channels, and a fresh look at targets based on unit economics.

Note: Log every significant change with the date and reason, in addition to change history. When performance shifts, you can quickly link cause and effect.

Scaling is where many accounts lose money, because extra spend usually buys conversions at a higher cost than the current average. The key concept is marginal CPA: the cost of the additional conversions you get from additional spend.

Worked example:

• Extra spend: ₹50,000. Extra conversions: 60.
• Marginal CPA = ₹50,000 ÷ 60 ≈ ₹833.
• If break-even CPA is ₹800, the additional ₹50,000 is losing money, even though the average CPA of ₹577 looks healthy.

How to scale sensibly:

1. Check headroom first. If a campaign is losing impression share to budget with a good CPA, raising the budget is the cheapest growth.
2. Loosen targets gradually. Raise target CPA or lower target ROAS in steps of about 10–20%, and let the system settle before the next step.
3. Expand reach deliberately: broad match with smart bidding, new keywords, new cities, new campaign types such as Demand Gen or Performance Max.
4. Use simulators for budget and targets to estimate the volume curve before committing.
5. Measure marginal returns after each step, not only the blended figure.

Efficiency versus growth: the right point depends on the business stage. A funded start-up chasing market share may accept a marginal CPA above first-order break-even if lifetime value justifies it; a profit-focused business should stop where marginal CPA meets its limit.

Note: Saying “I scale until marginal CPA reaches the break-even or LTV-based limit” is exactly the answer senior interviewers look for.

When CPCs rise suddenly, the cause is usually one of three things: more competition, lower ad quality, or a change you made. Auction Insights helps with the first and points you towards the others.

What Auction Insights shows for other advertisers in the same auctions:

• Impression share — how often each competitor showed.
• Overlap rate — how often their ad showed when yours did.
• Position above rate — how often their ad appeared above yours.
• Top of page rate and absolute top of page rate.
• Outranking share — how often you ranked higher or showed when they did not.

Diagnostic workflow for a CPC spike:

1. Confirm the scope. Is the rise account-wide or in specific campaigns, keywords or devices?
2. Check your own changes in the change history: new bid strategy, raised targets, new broad match keywords, or location changes.
3. Run Auction Insights for the affected period against the previous period. A new competitor with high overlap and a rising position-above rate is a common cause, for example a funded start-up launching in your category.
4. Check Quality Score and its components on key keywords. A drop in expected CTR or landing page experience raises CPC for the same position.
5. Look at seasonality and external events: festive sales, exam results, IPL season or budget-heavy quarter ends push CPCs up across the market.

Responses: improve relevance and landing pages, shift budget to less contested keywords and hours, strengthen brand defence if competitors bid on your name, and revisit targets if the market has permanently become more expensive.

Note: Auction Insights shows who is competing, not their bids or budgets. Avoid claiming it tells you what competitors pay.

Optimization Score is a percentage estimate, from 0% to 100%, of how well an account is set up according to Google’s recommendations. Each recommendation shows how much it would raise the score.

What it is and is not:

• It measures adoption of recommendations, not performance. An account at 100% can be unprofitable, and one at 70% can be excellent.
• Recommendations are generated automatically and often favour more reach and spend: raising budgets, adding broad match, switching bid strategies, or adding keywords.
• Dismissing a recommendation also counts towards the score, which shows that the score reflects review rather than blind acceptance.

How to handle recommendations:

• Usually worth applying: fixing disapproved ads, adding missing assets, fixing conversion tracking issues, and removing redundant or conflicting keywords.
• Evaluate carefully: budget increases, broad match expansion, and bid strategy changes. Check them against your CPA or ROAS targets and test with an experiment if the change is big.
• Often dismiss: suggestions that conflict with the business, such as keywords for services you do not offer or raising budgets for campaigns already at break-even.

Auto-apply recommendations: review which types are switched on. Some accounts have had keywords added or bid strategies changed without anyone noticing. Turn off anything that should be a deliberate decision.

Worked example: an account had a recommendation to raise a campaign’s budget by 40%. Marginal CPA analysis showed extra spend would come at ₹1,100 against a break-even of ₹900, so it was dismissed with a note.

Note: Say you review recommendations weekly and apply those that align with goals. That balanced view is what interviewers want.