Everything that actually matters in Google Ads and Meta Ads.
One manual, two ad platforms, no fluff. Structure, campaign types, bidding, targeting, tracking, creative, and the strategy that ties them together — written to be read once and returned to for years.
How the auction actually works
Google Ads is not a "highest bidder wins" system. Every time someone searches, an auction runs among every advertiser eligible for that query, and rank is decided by a formula, not a wallet.
01.1Ad Rank, in plain terms
Ad Rank = Your bid × Quality Score × the expected impact of your ad formats/extensions × a set of real-time auction-time signals (device, location, time of day, search intent). The advertiser with the highest Ad Rank wins the top position — and thanks to the second-price logic baked into the auction, you typically pay just enough to beat the Ad Rank of the advertiser below you, not your full maximum bid.
This is why a smaller budget with excellent Quality Score regularly beats a bigger budget with mediocre ads. It's the single most under-used lever in the entire platform.
01.2Quality Score — the three inputs
Expected CTR
How likely people are to click your ad for that keyword, relative to other advertisers on the same term.
Ad relevance
How closely your ad copy actually matches the intent behind the keyword — not just whether the word appears.
Landing page experience
Page speed, mobile-friendliness, and whether the page delivers what the ad promised.
Quality Score (shown 1–10) is a diagnostic, not a bidding lever — you can't target it directly, but every one of the three inputs above is something you can rewrite this week: tighter ad groups, copy that mirrors the keyword's intent, and a landing page that loads fast and matches the ad.
01.3Auction Insights report
This report shows how you stack up against other advertisers bidding on the same auctions: impression share, overlap rate, position-above rate, and top-of-page rate. It's the closest thing Google gives you to seeing your competitors' scoreboard — use it monthly to catch a competitor suddenly outbidding you before your traffic quietly drops.
Account structure
Everything in Google Ads nests inside a strict hierarchy. Get this shape right and every other setting — bidding, budgets, reporting — becomes simple. Get it wrong and you'll fight the account forever.
- Account — billing, users, linked assets (GA4, Merchant Center, YouTube channel), account-level negative keyword lists.
- Campaign — the level where budget, campaign type, bid strategy, network, location and language targeting live. Never mix goals inside one campaign; each one should have a single job.
- Ad group — a tight cluster of closely related keywords (for Search) or a set of assets (for Performance Max/Demand Gen), each pointing at one theme so your ad copy can stay relevant.
- Keywords / Ads / Assets — the actual bidding units and creative.
Campaign types
Each campaign type is really a different distribution channel with its own inventory, automation level, and creative requirements.
| Type | Where it shows | Best for | Control level |
|---|---|---|---|
| Search | Google search results, search partners | High-intent, "ready to act" queries | High — keywords, copy, extensions |
| Performance Max | Search, Display, YouTube, Discover, Gmail, Maps — all at once | Full-funnel automation off one goal + asset group | Low-medium — you feed assets and signals, AI does placement |
| Display | 2M+ sites/apps on the Google Display Network | Awareness, remarketing at scale, visual storytelling | Medium |
| Shopping | Search, Shopping tab, Images | E-commerce with a product feed | Medium — feed-driven, not keyword-driven |
| Video (YouTube) | YouTube in-stream, in-feed, Shorts | Awareness, consideration, video storytelling | Medium |
| Demand Gen | YouTube, Discover, Gmail, and — since the 2024 expansion — Instagram & Facebook feeds via Google Ads | Scroll-native, visual mid-funnel demand | Medium — image/video-led, lookalike-style audiences |
| App | Search, Display, YouTube, Play Store, Discover | App installs and in-app actions | Low — highly automated |
| Local | Search, Maps, Display, YouTube, Gmail | Driving visits to a physical location (great fit for showrooms) | Low — automated, feeds off your Business Profile |
03.1Performance Max, unpacked
PMax is a single campaign that bids across every Google inventory type at once, organised into asset groups (a bundle of headlines, images, videos, logos and a landing page URL — one per audience/theme) rather than ad groups. You still guide it with audience signals (your own customer lists, in-market segments) and, increasingly, search themes — but ultimately the algorithm chooses where and to whom your assets are shown. Two features advertisers asked for and got: channel-level reporting (to see the Search vs. Display vs. YouTube split) and brand exclusion controls.
A newer layer worth knowing: AI Max for Search campaigns, an opt-in feature inside standard Search campaigns that broadens keyword matching further using AI, automatically generates additional headlines/URLs, and expands final URLs — effectively bringing some of PMax's automation into keyword-based Search without leaving that campaign type.
03.2Demand Gen, explained
Renamed from "Discovery ads" in 2024, Demand Gen is built for scroll-native, visually led placements — feed and Reels-style content rather than search intent. Its most useful expansion: campaigns can now run creative across Instagram and Facebook placements directly from Google Ads, which is a genuine bridge between the two ecosystems this guide covers.
Keywords & match types
Match types decide how loosely or tightly a search query has to resemble your keyword before your ad is even allowed into the auction.
| Type | Syntax | Triggers on |
|---|---|---|
| Broad match | bike service | Any query Google's AI judges as related in meaning — widest reach, needs Smart Bidding to stay efficient |
| Phrase match | "bike service" | Queries that include that meaning/phrase, with words allowed before or after |
| Exact match | [bike service] | Queries that mean the same thing as the keyword — still allows close variants, no longer literally identical only |
Google's own current recommendation is broad match paired with Smart Bidding (Maximize Conversions/Value, or Target CPA/ROAS) rather than manually segmenting by match type — the bidding algorithm sets a bid for each individual auction based on how likely that specific query is to convert, so broad match plus automated bidding can out-learn a rigid exact-match-only structure. That said, broad match without a disciplined negative keyword list is the single fastest way to burn a small budget — treat the two as a pair, never one without the other.
04.1Negative keywords
Negatives block your ad from showing on irrelevant queries — "free," "jobs," "DIY," a competitor brand name, or a model your dealership doesn't carry. Build a shared negative list at the account level for evergreen exclusions (job seekers, DIY researchers) and add campaign-specific negatives from your Search Terms report weekly — this single habit is the highest-leverage 15 minutes in Google Ads account management.
Bidding strategies
Bidding splits into manual control and Smart Bidding (Google's machine-learning bid-setting, evaluated fresh in every single auction). As of June 2026, Google reorganised how these are labelled — worth knowing so old screenshots and new interface don't confuse you.
| Strategy | Optimises for | Use when |
|---|---|---|
| Maximize Clicks | Most clicks within budget | Brand-new campaign with zero conversion history, or a pure traffic/awareness goal — treat as a 2–4 week data-gathering phase, not a permanent strategy |
| Maximize Conversions | Most conversions within budget, no target | You have some conversion data but not enough volume for a reliable target yet |
| Target CPA | Hits a cost-per-conversion you set | Lead gen where every conversion is worth roughly the same — the standard choice for a dealership's "book a test ride" or "get a quote" goal |
| Maximize Conversion Value | Most total value within budget, no target | E-commerce with value data but still building volume |
| Target ROAS | Hits a return-on-ad-spend ratio you set | E-commerce/high-ticket sales where conversion values genuinely differ |
| Enhanced CPC | Adjusts your manual bids up/down per auction | You want to keep manual control but let Google nudge bids at the margins |
| Target Impression Share | Visibility at a chosen page position | Brand-defence campaigns — protecting your own branded terms from competitor bidding |
05.1Manual bidding
Manual CPC (with optional Enhanced CPC) lets you set the exact max bid per keyword or ad group. It's rare to see this outperform Smart Bidding once you have 30+ conversions a month, but it's still useful for brand-new accounts, tightly capped test budgets, or keywords you want to bid on for strategic reasons regardless of predicted ROI.
05.2Portfolio bid strategies
A portfolio bid strategy groups several campaigns under one shared Target CPA/ROAS (or Maximize Conversions/Value) model instead of each campaign learning in isolation. The pooled conversion data lets Smart Bidding learn faster and generalise better — advertisers commonly report meaningfully higher account-level ROAS after consolidating from campaign-level to portfolio strategies, though the exact lift depends heavily on how fragmented the account was beforehand. Group campaigns that share a genuine business goal (e.g. every brand's "Service Booking" campaign across all cities), not campaigns that merely use the same bid strategy by coincidence — mixing unrelated goals under one shared target usually makes both perform worse, not better.
05.3Seasonality adjustments vs. Promotion Mode
These solve different problems and are easy to conflate:
Seasonality adjustments
You tell Smart Bidding "expect conversion rate to rise/fall by X% for these dates" (ideal for 1–7 day events; use with caution beyond 14 days). Bids adjust in expectation of the change, then return to baseline automatically once the window ends — no manual reversal needed. Available for Search, Shopping and Display on Target CPA/ROAS, and for Performance Max and App campaigns on any bid strategy.
Promotion Mode (beta)
A newer, separate tool (Search and Performance Max only at beta launch) that schedules a temporary change to your ROAS tolerance plus extra daily budget across a defined peak window — built for flash sales and launches where you deliberately want the algorithm to spend more even at a temporarily lower ROAS, not just predict a conversion-rate shift.
05.4What bid adjustments still do under Smart Bidding
Once a campaign is on Smart Bidding, most manual bid adjustments are simply ignored — the algorithm re-derives its own per-auction bid from the target you set. A short list of exceptions still functions:
- Device exclusion at −100% — the one device adjustment that works across every automated strategy, since it's a hard exclusion, not a bid nudge.
- Device modifiers on Target CPA — these adjust the CPA target itself for that device rather than the bid, so they still have an effect, just not the one manual bidders expect.
- Seasonality adjustments and data exclusions — both explicitly designed to feed Smart Bidding extra context, so they remain fully functional (a data exclusion matters most after a tracking outage, so the bad-data window doesn't quietly poison the model).
- Ad schedule (dayparting) bid adjustments — still work, but only under Maximize Clicks; they're not available for App campaigns at all, and other Smart Bidding strategies won't apply them.
Everything else — location, audience, and generic device bid modifiers on Target CPA/ROAS/Maximize Conversions/Maximize Conversion Value — is effectively cosmetic once Smart Bidding is live. Treat quarterly review as cleanup of legacy modifiers that no longer do anything, not as an active lever.
Ad formats & assets (extensions)
"Extensions" were renamed "Assets" — same idea: extra pieces of information attached to your ad that make it bigger, more useful, and more clickable, at no extra cost per click.
06.1Responsive Search Ads (RSAs)
The only Search ad format now available: you supply up to 15 headlines and 4 descriptions, Google's AI tests combinations and shows the best-performing mix per auction. Write each headline so it can stand alone (Google will mix and match them), pin only when brand/legal compliance truly requires it — pinning too much defeats the purpose of the format.
06.2Key assets to always fill in
- Sitelinks — extra links under the ad (e.g. "Book a Test Ride," "EMI Calculator," "Service Booking")
- Callouts — short non-clickable trust phrases ("0% Down Payment," "5 Showrooms in Lucknow")
- Structured snippets — categorised lists ("Models: Classic 350, Hunter 350, Meteor 350")
- Call assets — a clickable phone number, essential for dealership walk-in/call conversions
- Location assets — pulls from your Google Business Profile, shows address and distance
- Lead form assets — an in-ad form for Search/Display, no landing page required
- Price assets — a small pricing carousel under the ad
Audiences & targeting
Google's audience layer has quietly shifted from "manually pick a segment" toward "give the algorithm a seed and let it expand" — mirroring the direction of the whole platform.
| Type | What it is |
|---|---|
| Remarketing / RLSA | People who already visited your site or app — your highest-intent, cheapest-to-convert pool |
| Customer Match | Upload your own customer/lead list (email, phone) to target or exclude — direct bridge to your CRM's contact list |
| In-market segments | People actively researching/comparing products like yours right now |
| Affinity segments | Longer-term lifestyle and interest patterns |
| Detailed demographics | Life events and demographic slices (e.g. new homeowners, recently married) |
| Optimized Targeting | Google's automated audience expansion beyond what you've manually selected, when it predicts better performance — this replaced the old "Similar Audiences" feature, which Google fully retired in 2023 |
Conversion tracking & attribution
Bidding is only as good as the conversion data feeding it. Get this wrong and every Smart Bidding strategy above is optimising toward noise.
08.1The stack
- Google tag — the base site-wide tag (replaces the old separate remarketing + conversion tags)
- GA4 linking — import Google Analytics goals as Google Ads conversion actions
- Enhanced Conversions — sends hashed first-party data (email/phone) alongside a conversion to recover data lost to browser privacy changes, meaningfully improving Smart Bidding accuracy
- Offline conversion import — upload conversions that happen after the click, off-platform (a showroom visit that becomes a sale three weeks later, tracked via GCLID)
08.2Attribution
Data-Driven Attribution (DDA) is now the default and, for most advertisers, effectively the only meaningful model — Google has been retiring the older rule-based models (first-click, linear, time-decay, position-based) for standard conversion actions. DDA uses your own account's conversion patterns to credit each ad interaction along the path, rather than crediting 100% to the last click.
Budgets, Optimization Score & account tools
09.1Budgets
Set at the campaign level as an average daily amount — Google can spend up to roughly double on a strong day and less on a slow one, balancing out over the month. Shared budgets pool spend across multiple campaigns; portfolio bid strategies go further and let Smart Bidding optimise across several campaigns toward one shared target.
09.2Optimization Score
A 0–100% score with recommendations attached. Useful as a checklist, dangerous as a target — some recommendations (broader match types, higher budgets) genuinely help; others exist to increase your spend. Read every recommendation before applying it; never bulk-apply.
09.3Tools worth knowing
- Google Ads Editor — free desktop app for fast bulk edits, offline drafting, and copy/paste between campaigns
- Experiments / Drafts — proper A/B tests that split traffic, rather than just eyeballing before/after performance
- Scripts — JavaScript automation for anything the interface can't do on a schedule (pausing keywords under a CPA threshold, alerting on budget pacing)
- Automated rules — simpler, no-code version of the above for common if/then actions
Local Services Ads
A separate product from standard Google Ads, built specifically for local, service-based businesses.
LSAs show above regular Search ads with a "Google Screened/Guaranteed" badge, and — unlike everything else in this section — you pay per lead, not per click. Leads come in as phone calls or messages, and you can dispute ones that are clearly invalid (wrong service, wrong area, spam). It requires a background/licence check to qualify, and coverage is limited to specific service categories and countries, so check current eligibility before assuming a dealership service department qualifies in your market.
Google Ads vs. Meta Ads: intent vs. interest
The single sentence that explains almost every strategic decision between these two platforms:
| Google Ads | Meta Ads | |
|---|---|---|
| Targeting logic | What someone is actively searching for | Who someone is / what they've engaged with |
| Best funnel stage | Mid-to-bottom — capturing existing intent | Top-to-mid — creating and nurturing intent |
| Typical cost pattern | Higher CPC, but the click already has intent behind it | Lower CPM, but you're paying to build intent, not just capture it |
| Creative demand | Lower — copy and offer matter more than visuals | Very high — creative quality is the #1 performance lever |
| Where it struggles | New products/categories nobody searches for yet | High-intent "I need this today" moments |
A full-funnel playbook combining both
The mistake most accounts make: running both platforms as two separate, uncoordinated silos, each measured against its own last-click metrics. A cleaner model treats them as one funnel:
- Meta builds the audience Google will later capture. Awareness and Engagement campaigns introduce the brand/model to cold audiences who then — days or weeks later — go and search for it by name.
- Google captures the resulting search demand, including branded search volume Meta helped create — this is why branded search often quietly rises when Meta spend increases, even with zero direct attribution link between the two.
- Both platforms retarget the same visitor pool — anyone who visited your site from either channel becomes a remarketing audience on both, since intent doesn't care which platform first brought them to you.
- Feed every closed sale back into both platforms as Customer Match (Google) and Custom Audiences (Meta) — this is the single highest-leverage data loop most small teams skip, and it directly improves both platforms' targeting over time.
Running both on a tight budget
With limited spend, the goal shifts from "cover every stage of the funnel" to "protect the highest-certainty conversions first, then expand."
- Retargeting first, always. A remarketing/Custom Audience segment is cheaper to convert than any cold audience on either platform — fund it fully before spending a rupee on cold prospecting.
- Pick one lead platform primarily, based on your buyer, not your preference. High-intent, ready-to-buy searches → weight Google Search + Local. Visual, consideration-heavy purchases, or a market where WhatsApp is the default channel → weight Meta + CTWA.
- Use automation to substitute for headcount, not to replace judgement. Smart Bidding and Advantage+ exist precisely so a small, lean team can run what used to require a bigger media-buying desk — but someone still has to check the negative keyword list and read the creative reports weekly.
- Let negatives and exclusions do the saving, not lower bids. A tight negative keyword list on Google and firm audience exclusions on Meta (existing owners, irrelevant geographies) recover more wasted budget than shaving bids ever will.
- Consolidate rather than fragment. Five ad sets each with ₹500/day teach the algorithm nothing; one ad set with ₹2,500/day exits the learning phase properly. Both platforms punish thin budgets spread across too many active campaigns.
- Use free/native lead capture before building landing pages. Google Lead Form assets, Meta Instant Forms, and Click-to-WhatsApp all remove the cost and time of a dedicated landing page — a real saving when budget covers media, not development.
- Measure cost-per-sale, not cost-per-click. A channel with a higher CPC but a much higher close rate is cheaper in practice — track cost through to an actual sale in your CRM, not just the platform-reported cost per result.
Playbook for a local, multi-location dealership
- Structure by brand + location, not by generic campaign themes — Bajaj/Royal Enfield/Tata CV each have different buyers, price points and sales cycles, and each showroom needs its own attributable performance data.
- Google Local campaigns + Local Services Ads for "near me" and walk-in intent, tied tightly to each location's Google Business Profile — accurate NAP (name/address/phone) consistency across every listing matters more here than almost anywhere else.
- Click-to-WhatsApp as the primary Meta lead format, not Instant Forms — a WhatsApp conversation lets your sales team answer the two or three questions that actually close a vehicle sale (on-road price, EMI, stock availability) faster than any web form.
- Feed every showroom visit and sale back into both platforms via Customer Match and Custom Audiences — this is the highest-leverage habit for a lean, self-managed multi-location account and the thing an agency would charge the most to maintain manually.
- Use Dynamic/Catalog ads for vehicle inventory once your product feed is clean, so someone who viewed a specific model sees that model again in remarketing, not a generic dealership ad.
Policy & compliance
Most disapprovals aren't mysterious once you see the pattern: both platforms are optimising for "would a stranger feel misled or targeted by this?" — not for whether your product is legitimate. Know the common triggers and you'll spend far less time in appeal queues.
05.1Google Ads — common disapproval categories
| Category | Typical trigger | Fix |
|---|---|---|
| Misleading claims | Implied promises, omitted material info, "results not typical" framing — Google's definition covers implication, not just outright falsehood | State only what you can substantiate; move conditions out of fine print and into the headline claim itself |
| Destination mismatch | Ad promises one thing, landing page delivers another (or the offer/price shown differs) | Keep ad copy and landing page price/offer identical, word for word where it matters |
| Editorial & format | Excessive punctuation/symbols, ALL CAPS, unclear grammar, broken headline/description assembly in RSAs | Write clean, sentence-case copy; preview all RSA combinations before publishing |
| Restricted content | Financial services, healthcare and a few other categories require identity/KYC verification before ads run at all | Complete the relevant Google advertiser verification programme ahead of a campaign launch, not during one |
| Trademark/copyright | Competitor brand names or logos used in ad creative or copy (bidding on them as keywords is fine — using them in the ad itself usually isn't) | Keep competitor names out of headlines/descriptions and creative assets |
| Technical/landing page | Broken links, non-crawlable pages (robots.txt blocking Googlebot), missing SSL, intrusive pop-ups/auto-downloads | Test the exact final URL in an incognito window before submitting; keep AdsBot-Google unblocked |
05.2Meta Ads — common disapproval categories
| Category | Typical trigger | Fix |
|---|---|---|
| Personal attributes | Copy that implies you know something about the individual viewer — age, health, financial status, relationship status — even indirectly ("Struggling to afford your next bike?") | Speak to the outcome, not the viewer's assumed condition: "Flexible financing on Royal Enfield models" instead of assuming financial hardship |
| Misleading claims | Exaggerated or unverifiable results; Meta doesn't require intent to deceive, only that a reasonable viewer could be misled | Drop superlatives you can't back up ("best," "guaranteed," "India's #1") unless independently substantiated |
| Discriminatory targeting | Excluding or including audiences based on protected attributes in ad-set targeting | Target on interest/behaviour signals, not protected-class proxies |
| Non-functional landing experience | Pages that auto-download files, trap navigation, or don't match the ad's stated destination | Land people on a real, navigable page that matches the ad |
| Account-level risk signals | Aggressive budget scaling, rapid new-account spend jumps, or a burst of near-identical ads can trigger elevated scrutiny independent of any single ad's content | Scale budgets gradually; avoid duplicating the same creative across many campaigns at once |
05.3India-specific: ASCI and RBI-adjacent rules for auto + EMI/finance copy
Neither Google nor Meta enforces Indian advertising law directly, but both will pull an ad reported for violating it, and the legal exposure sits with the dealership regardless of which platform served the ad. Two bodies matter most for a vehicle dealership running finance-led offers.
ASCI Code basics
India's self-regulatory advertising body (its Code has statutory backing under the Cable TV Networks Rules). Core rule: ads must not distort facts or mislead by implication or omission. A disclaimer can clarify a claim but can't contradict or override the main message.
Automotive-specific precedent
ASCI has previously upheld complaints against dealership/manufacturer ads showing stunts or unsafe riding in ordinary traffic — even with a disclaimer — because the visuals themselves promoted a hazardous practice. Keep vehicle demo footage to controlled settings, and disclose it clearly on-screen if you show performance riding.
RBI Digital Lending Directions
Governs how loan/EMI offers can be advertised and disclosed, even when the dealership isn't the lender — the rules bind the regulated entity (bank/NBFC) and any platform presenting their offers.
"No-cost EMI" is a specific trap
RBI guidance is explicit that EMI conversions with an interest component cannot be camouflaged as zero-interest/no-cost — the interest is typically folded into the sticker price instead. Advertising "0% EMI" without knowing exactly how your financing partner structures it risks a misleading-claim complaint on both the ASCI and platform side.
- Never quote a specific interest rate, EMI amount or "starting from" figure in ad copy unless it's the actual APR (interest + processing fees + any other charge, not just the headline interest rate) currently offered by the named financing partner — RBI's Key Fact Statement (KFS) framework requires the all-inclusive cost to be disclosed to the borrower, and platform reviewers increasingly treat a headline rate that omits fees as a misleading claim.
- Name the financing partner, don't imply the dealership is the lender. "EMI options available through our partner banks/NBFCs" is accurate; "Get your bike for ₹2,999/month" implies a fixed, guaranteed offer the dealership itself may not be able to honour once a specific buyer's credit profile is assessed.
- Avoid guaranteed-approval language. "Instant approval," "no rejection," "loan guaranteed" are the kind of claims both ASCI and RBI-adjacent scrutiny target directly, since approval always depends on the lender's underwriting.
- Keep safety disclaimers legible, not decorative. ASCI's disclaimer guidance specifically penalises fast, quiet, or visually crowded disclaimers that a viewer can't actually absorb — if a video ad shows any stunt riding, the safety disclaimer needs to be readable at normal viewing speed, not a blink-and-miss frame.
Troubleshooting: symptom → likely cause
Almost every underperforming campaign fits one of these patterns. Diagnose in this order — funnel stage by funnel stage — before touching bids or budget, since bid changes rarely fix a creative or landing-page problem.
| Symptom | Likely cause | Where to look / fix |
|---|---|---|
| High CPM + low CTR | Creative problem — weak hook, stale creative, or wrong format for the placement | Refresh creative first; check frequency (Meta) for fatigue before assuming targeting is wrong |
| Good CTR + no/low conversions | Offer or landing-page problem — the click delivered, the destination didn't | Check page load speed, offer clarity, and that the tracked conversion event actually fires |
| Good CTR + high cost-per-lead | Landing-page friction — form too long, no trust signals, no clear next step | Cut form fields to the minimum needed to qualify a lead; add a phone/WhatsApp option |
| Low CTR + low CPM (impressions are cheap, nobody bites) | Targeting/relevance mismatch — reaching people with no interest in the offer | Review Search Terms report (Google) or interest/placement breakdown (Meta); tighten |
| Conversions tracked, but sales don't materialise in the showroom | Lead-quality problem, not a tracking problem — junk fills, bot submissions, or overly broad reach | Add a qualifying question to the form; exclude low-quality placements/apps; check the CRM's lead-to-sale rate by campaign, not the platform's reported CPL |
| Sudden drop in impressions/delivery | Learning phase reset from a recent edit, a bid/budget set too low to compete, or a "Limited" status from policy/approval | Check the change history for recent edits and the account-level status banner before assuming demand dried up |
| Rising CPM + rising frequency (Meta) | Audience fatigue — the pool is too small for the budget/duration | Expand the audience, add fresh creative, or move budget toward a broader Advantage+ set |
| Search impression share lost to budget | Budget capped below what demand supports | Raise the campaign budget or reallocate from a lower-performing campaign |
| Search impression share lost to rank | Quality Score / Ad Rank problem, not a budget problem | Rewrite ad copy for tighter keyword relevance and check landing page experience |
| Good volume, poor ROAS | Wrong bid strategy for the goal, or conversion value not passed correctly to the platform | Verify the actual sale value is flowing into the conversion action, not a placeholder or lead-only value |
UTM & campaign-naming convention
A naming system only earns its keep if every campaign, across every brand and location, can be filtered and rolled up consistently in GA4 or a CRM without manual cleanup. Designed here for three brands (Bajaj, Royal Enfield, Tata Motors CV) across 22 locations.
07.1Campaign-name pattern (in-platform, not the URL)
{brand}_{city}_{objective}_{funnel}_{yymm}
Example: re_lko_leadgen_bofu_2607 — Royal Enfield, Lucknow, lead-generation objective, bottom-of-funnel, built July 2026. Lower-case, underscore-separated, no spaces — this survives being pasted into a spreadsheet, a URL, or a script without re-encoding.
| Segment | Codes |
|---|---|
| Brand | bjj Bajaj 3-wheelers · re Royal Enfield · tcv Tata Commercial Vehicles |
| City (sample — assign a fixed 3-letter code per hub city, reuse everywhere) | lko Lucknow · knp Kanpur · brl Bareilly · gkp Gorakhpur · … one code per showroom city, kept in a single reference sheet so it never drifts |
| Objective | leadgen · awareness · traffic · retarget · catalog |
| Funnel stage | tofu · mofu · bofu |
07.2UTM parameters (the URL layer)
| Parameter | Value pattern | Example |
|---|---|---|
| utm_source | Platform | google / meta / whatsapp |
| utm_medium | Paid channel type | cpc / paid-social / ctwa |
| utm_campaign | Mirrors the in-platform campaign name exactly | re_lko_leadgen_bofu_2607 |
| utm_content | Creative/ad variant identifier — lets you A/B compare inside one campaign | carousel-a / video-hook2 |
| utm_term | Keyword (Google Search only) or audience label (Meta) — optional, only when it adds real filtering value | re-hunter-350-lucknow |
A/B test design template
A real test isolates one variable, runs long enough to clear both platforms' learning phase, and ends with a predetermined decision rule — not a vibes-based read of whichever variant "felt" like it was winning after three days.
08.1Platform tools
Google — Drafts & Experiments
Create a Draft (a sandboxed copy of a live campaign), edit the one variable you're testing, then launch it as an Experiment with a defined traffic split and end date. Google splits traffic and reports statistical significance directly.
Google — Performance Max experiments
Newer PMax campaigns support experiment comparisons against a standard Search campaign or another PMax setup, useful for the "should this go into PMax at all" question specifically.
Meta — Experiments (A/B Test)
Ads Manager's Experiments tool randomly and evenly splits a defined audience between ad sets so each variant competes for genuinely comparable people, rather than letting delivery optimisation quietly favour one.
Meta — Conversion Lift / holdout test
A separate incrementality tool: it withholds ads from a random holdout group entirely, then compares their conversion rate to the exposed group — answering "did this campaign cause incremental sales" rather than just "which variant had a better CTR."
08.2Test design template
| Field | What to fill in |
|---|---|
| Hypothesis | One sentence: "Changing X will improve Y because Z." Forces you to state the mechanism, not just the metric. |
| Variable tested | Exactly one — creative hook, headline, audience definition, bid strategy, landing page, or offer. Never two at once. |
| Control | The current, unchanged version |
| Variant(s) | The single-variable change, described precisely enough that someone else could rebuild it |
| Primary metric | The one number that decides the winner — not a scoreboard of five metrics pointing different directions |
| Minimum duration | At least one full purchase/consideration cycle, and long enough to clear the learning phase (typically 1–2 weeks minimum on Meta after any significant edit; Google Experiments are commonly run 4+ weeks for statistical confidence on lower-volume accounts) |
| Minimum sample/spend | Enough conversions per arm for the result to be meaningful — as a rough floor, aim for at least 30–50 conversions per variant before reading the result as reliable |
| Decision rule | Written before the test starts: "If Variant B beats Control by more than X% on the primary metric with the platform reporting statistical significance, ship B." |
Feed setup mechanics: vehicle inventory
Both platforms' vehicle ad formats are entirely feed-driven — there's no custom ad copy to write. Feed quality is the single biggest lever on whether these formats work at all.
09.1Google Merchant Center — vehicle feed
- Required attributes: VIN (or a unique ID where VIN isn't applicable), make, model, year, condition (new/used), price, availability, and a URL to the vehicle's detail page (VDP).
- Recommended attributes that improve matching: mileage, exterior/interior colour, trim level, transmission, fuel type, drivetrain, body style, and multiple high-quality images (Google recommends roughly 1200×628 for landscape framing).
- VDP requirements: the linked landing page must independently display dealership name, location, price, VIN, mileage (for used vehicles) and availability — Google checks the page itself, not just the feed.
- Refresh cadence: update the feed at least daily, ideally every few hours; sold vehicles need to drop out immediately, and most implementations expire a listing automatically after a few days without a refresh.
- Format: CSV (optionally ZIP/GZ-compressed), submitted as a primary data source in Merchant Center with Google's exact attribute headers.
09.2Meta Catalog — Automotive Inventory Ads
- Catalog type: create it as an Auto catalog specifically (Commerce Manager → Add Catalogue → Auto), not a generic e-commerce catalog — this unlocks the vehicle-specific fields and ad templates.
- Core required fields (Meta's automotive template has around 20 required columns out of ~80 total): unique vehicle ID, title, description, VDP URL, image link, price, make, model, year, condition/state of vehicle (new/used), VIN, body style, mileage (value + unit), and availability.
- Common recommended fields: exterior colour, interior colour, transmission, fuel type, drivetrain, trim, dealer name and address (for local-inventory matching).
- Upload path: Commerce Manager → Catalog → Data sources → Add items → Data feed → upload the CSV/TSV/XML (or connect a scheduled feed URL) → confirm currency and column mapping.
- Refresh cadence: schedule the feed fetch at least every 24 hours; for fast-moving inventory, more frequent is safer against advertising a sold vehicle.
- Pixel/CAPI connection: link the Meta Pixel and/or Conversions API to the catalog so Automotive Inventory Ads can retarget people who viewed a specific vehicle with that same vehicle, not a generic dealership ad.
09.3The sync process, end to end
- Source of truth — your DMS or inventory sheet is the only place vehicle data gets edited; the feed is generated from it, never hand-edited separately.
- Feed generator — a script (Apps Script, or any scheduled job) transforms the source data into the exact column headers each platform requires, since Google's and Meta's attribute names don't match each other and can't share one raw file unmodified.
- Hosted feed URL — the generated CSV/XML is hosted somewhere both platforms can fetch it on a schedule (a public but unlisted URL, or an authenticated feed endpoint).
- Scheduled fetch — both Merchant Center and Commerce Manager are pointed at that URL with a fetch schedule matching your refresh cadence.
- Diagnostics check — after every feed change, check Merchant Center Diagnostics and Commerce Manager's catalog error report before assuming the update went live cleanly; a single malformed row can silently drop vehicles from eligibility.
New campaign launch checklist
Tick items off before you hit publish. State resets on page reload — nothing is saved to this device or anywhere else, so treat it as a per-session pre-flight check, not a permanent record.
Glossary, A–Z
Every acronym and term used above, in one searchable list. Colour dot shows which platform(s) it applies to.
- Ad Rank
- Google's formula deciding auction position: bid × Quality Score × expected format impact × auction-time signals.
- Advantage+
- Meta's automation suite spanning audiences, placements, budgets and creative.
- AEM
- Aggregated Event Measurement — Meta's privacy-preserving iOS measurement framework.
- AIA
- Automotive Inventory Ads — Meta's dynamic ad format built entirely from a vehicle catalog feed.
- AI Max for Search
- Opt-in Google feature broadening keyword matching and auto-generating assets inside Search campaigns.
- Andromeda
- Meta's ad-ranking retrieval engine, in production since 2024.
- ASCI
- Advertising Standards Council of India — the self-regulatory body reviewing ad honesty and decency; its Code carries statutory weight under the Cable TV Networks Rules.
- ABO
- Ad Set Budget Optimization — fixed budget set per ad set rather than pooled at campaign level.
- Broad match
- Google keyword match type triggering on any query judged related in meaning.
- CAPI
- Conversions API — server-side event tracking, sent directly from your backend to Meta or Google.
- CBO
- Campaign Budget Optimization — one shared budget Meta allocates across ad sets automatically.
- CPA
- Cost Per Acquisition — total spend divided by conversions.
- CPC
- Cost Per Click.
- CPL
- Cost Per Lead.
- CPM
- Cost Per Mille — cost per 1,000 impressions.
- CTWA
- Click-to-WhatsApp Ads — an ad whose CTA opens a WhatsApp chat.
- CTR
- Click-Through Rate — clicks divided by impressions.
- ctwa_clid
- The click ID Meta passes into a WhatsApp conversation, required in CAPI events to attribute a conversation back to its ad.
- Custom Audience
- A Meta audience built from your own data — customer lists, site visitors, engagers.
- Customer Match
- Google's equivalent of Custom Audiences — upload your own contact list to target or exclude.
- DDA
- Data-Driven Attribution — Google's default attribution model, crediting ad interactions along the path using your own account's conversion patterns.
- Dynamic/Catalog Ads
- Ads auto-assembled per viewer from a connected product catalogue.
- Enhanced Conversions
- Google feature sending hashed first-party data alongside a conversion to recover data lost to browser privacy limits.
- Estimated Action Rate
- Meta's per-user prediction of how likely someone is to take your optimisation event — one of the three inputs into the auction's Total Value score.
- Exact match
- Google keyword match type triggering on queries meaning the same thing as the keyword.
- Frequency
- Average number of times a unique person saw your ad.
- GCLID
- Google Click ID — used to tie offline conversion imports back to the originating ad click.
- GMC
- Google Merchant Center — hosts the product/vehicle feeds that power Shopping and Vehicle Ads.
- GSP auction
- Generalized Second-Price auction — Google typically charges just enough to beat the bidder ranked below you, not your full max bid.
- KFS
- Key Fact Statement — RBI-mandated disclosure of a loan's all-inclusive APR, fees and tenure, shown to a borrower before disbursal.
- LSA
- Local Services Ads — pay-per-lead Google product for local service businesses, badged "Google Screened/Guaranteed."
- Lookalike Audience
- A Meta audience of new people statistically similar to a source audience you provide.
- Learning phase
- The early period after launch or a major edit where both platforms' algorithms are still gathering data — significant changes during this window reset it.
- Negative keyword
- A term that blocks your ad from showing on matching searches.
- ODAX
- Outcome-Driven Ad Experiences — Meta's 6-objective campaign framework.
- Optimization Score
- Google's 0–100% account health/recommendation score — a checklist, not a target.
- Optimized Targeting
- Google's automated audience expansion beyond manually selected segments; replaced the retired "Similar Audiences."
- Performance Max (PMax)
- A single Google campaign bidding across Search, Display, YouTube, Discover, Gmail and Maps at once.
- Phrase match
- Google keyword match type triggering on queries containing that phrase's meaning, with extra words allowed around it.
- Pixel
- Meta's client-side JavaScript tracking tag.
- Portfolio bid strategy
- A shared Target CPA/ROAS (or Maximize Conversions/Value) model applied across several campaigns at once so they learn from pooled conversion data instead of in isolation.
- Promotion Mode
- Beta Google Ads tool (Search and PMax) that schedules a temporary ROAS-tolerance change plus extra daily budget across a defined peak window, distinct from seasonality adjustments.
- Quality Score
- Google's 1–10 diagnostic of expected CTR, ad relevance and landing page experience.
- Reach
- Unique people who saw your ad, as distinct from impressions (total views, including repeats).
- Remarketing / RLSA
- Targeting people who already visited your site or app; RLSA applies this to Google Search specifically.
- ROAS
- Return On Ad Spend — revenue generated divided by ad spend.
- RSA
- Responsive Search Ad — the current Google Search ad format, testing combinations of up to 15 headlines and 4 descriptions.
- Search Terms report
- Shows the actual queries that triggered your ads — the primary source for new negative keywords.
- Seasonality adjustment
- A scheduled signal telling Smart Bidding to expect a conversion-rate change for a short (1–7 day) upcoming event; bids return to baseline automatically afterward.
- Smart Bidding
- Google's umbrella term for machine-learning bid strategies optimising at auction time: Target CPA, Target ROAS, Maximize Conversions, Maximize Conversion Value.
- Target CPA / Target ROAS
- Smart Bidding strategies aiming for a set cost-per-conversion or return-on-ad-spend; restored as standalone-named strategies in Google's June 2026 relabel.
- TOFU / MOFU / BOFU
- Top/Middle/Bottom of Funnel — shorthand for a buyer's distance from purchase.
- Total Value
- Meta's auction score deciding which ad wins an impression — roughly bid × estimated action rate + ad quality — not simply the highest bidder.
- UTM parameters
- Tags appended to a URL to track a click's source/medium/campaign in analytics tools.
- VDP
- Vehicle Detail Page — the landing page a vehicle ad or catalog listing links to; must independently show price, VIN, mileage and availability to stay policy-eligible.
- Value Rules
- Meta bid multipliers, layered on ROAS-goal bidding, that raise or lower the effective value assigned to conversions from a chosen audience segment.