Counting and capture come first
Fixing the count will drop reported conversions from 190 towards 150. That changes the numbers, not the demand, so don’t read it as a performance fall.
Guide · Google Ads audit
Audit the evidence chain before the settings. Ask what changed, whether Google can see the leads that matter, and whether spend goes to searches that produce them. Only then touch keywords, bids and budget.
Where to start
Short answer
Start with what changed: change history, auto-applied recommendations and automatic upgrades such as AI Max. Then check that each lead is counted once and that Google can see which leads sales qualified. Search terms, bids and budgets come after that, because every later finding depends on trustworthy conversion data.
Most audit checklists list 50 or 100 settings with no order and no way to decide what to fix first. In a B2B account that is dangerous. If the conversion count is wrong, a search terms review, a bid strategy review and a budget review all reach confident, wrong conclusions. The ladder below puts the checks in the order their evidence depends on each other, and stops you at the first rung that fails.
Symptom router
An audit usually starts because something feels wrong. Use the symptom to decide which rungs to read first, then work through the rest.
| Symptom | Rungs to read first | Also read |
|---|---|---|
| No leads, or far fewer than usual | 0, 1, 6, 8 | – |
| Leads arrive, but sales says they’re junk | 2, 3, 7 | Junk leads from ads |
| Cost per lead rising | 0, 4, 6 | – |
| Performance dropped straight after a change | 0 | – |
| Nobody can tell what’s working | 1, 2 | Paid media pipeline attribution |
Original framework
Nine rungs. Each has a gate question, a place to look and a red flag. When a rung fails, fix it before trusting anything on the rungs above.
| Rung | Gate question | Where to look | Red flag | Deeper guide |
|---|---|---|---|---|
| 0 | What changed? | Change history, auto-applied recommendations, AI Max settings | A performance shift within a week of a change nobody owns | This page |
| 1 | Does Google count real leads, once? | Goals summary: primary and secondary actions, counting setting | Google conversions above CRM ad leads by more than a small, explained margin | Conversion tracking |
| 2 | Can Google see which leads were qualified? | Qualified-lead import, upload errors, click ID and email capture | No qualified-lead action, or most ad leads carry no click ID or email | Offline conversion import |
| 3 | Is spend going to buyer intent? | Search terms for Search, AI Max and PMax, by qualified-lead action | Term groups with spend and no qualified leads over a meaningful number of leads | B2B negative keywords |
| 4 | Do campaigns compete or leak? | Exact match coverage, PMax search terms, channel performance report | Your exact or brand terms showing in PMax search terms | Search vs Performance Max |
| 5 | Is bidding learning from the right goal? | Bid strategy, its conversion goal and status | Bidding on form fills while sales rejects most of them | Smart Bidding for lead generation |
| 6 | Is budget the real constraint? | Search lost impression share (budget) and (rank), limited by budget status | High lost share to budget on a campaign with a good cost per qualified lead | B2B Google Ads budget |
| 7 | Do ads and pages qualify the click? | Message match, form fields, spam controls, landing pages report | Name-and-email forms; clicks landing on pages without a form | Google Ads lead quality |
| 8 | Are targeting settings deliberate? | Location option, language, search partners, ad schedule | Defaults left on that don’t fit where you sell | GCC B2B paid ads |
Change history covers the last two years, and you can undo most changes made in the last 30 days (Google Ads Help). Check auto-applied recommendations from the Recommendations page, where the history shows what was applied (Google Ads Help). Google is auto-upgrading Search campaigns that use automatically created assets or the campaign-level broad match setting to AI Max from September 2026, with Dynamic Search Ads following from February 2027 (Google). And since 17 August 2026, budget-limited campaigns on a target-based bid strategy perform more consistently towards their target (Google Ads Help), so a campaign that used to beat its target may not any more.
For leads, Google describes the “One” counting setting, one conversion per ad click, as the good choice (Google Ads Help). Phone-click and WhatsApp-click actions set as primary inflate the count without proving an enquiry. If you use enhanced conversions for leads, its diagnostics report grades set-up health as Excellent, Good, Needs attention or Urgent (Google Ads Help).
Google suggests bidding on an action with at least 15 conversions in the last 30 days at account level, ideally one that happens within 7 days of the ad interaction (Google Ads Help). Since 15 June 2026, API uploads of offline conversions and enhanced conversions for leads go through the Data Manager API (Google Ads Help): an old integration that stopped uploading is a rung 2 failure.
Read the AI Max match type in Search and the search terms report now available for all Performance Max campaigns (Google Ads Help). A Search keyword that exactly matches the query is preferred over Performance Max (Google Ads Help), but a budget-limited campaign won’t always serve, which is how PMax picks up exact terms (Google Ads Help).
Target CPA can start with no conversion history; Google recommends evaluating it over the last 30 days with at least 30 conversions (Google Ads Help). Changing a target doesn’t trigger learning (Google Ads Help); a new strategy, a setting change or adding and removing campaigns does (Google Ads Help). “Revert to Maximize Clicks to gather data” is rarely the fix in B2B: the conversion choice usually is.
Search lost impression share (budget) is the share of time ads didn’t show because of budget; (rank) is the share lost to Ad Rank (Google Ads Help). For forms, Google’s lead generation guidance names qualifying questions, reCAPTCHA and server-side validation (Google Ads Help). Treat Quality Score as Google does: a diagnostic, not a KPI and not an auction input (Google Ads Help). Location targeting defaults to “Presence or interest”, which also reaches people interested in your locations (Google Ads Help).
The free ad account assessment takes 10–15 minutes, needs no CRM and makes no campaign changes during the assessment.
Reconciliation
Short answer
Line up, for each campaign and the same date range, Google’s primary conversions, the ad leads in your CRM or sheet, how many carry a click ID or email, and how many sales qualified. Large gaps usually mean duplicate counting, the wrong primary action, or leads arriving by phone or WhatsApp.
| Campaign | Google primary conversions | CRM ad leads | With click ID or email | Qualified | Qualified imported to Google | Quotes or opportunities | Cost per qualified lead |
|---|---|---|---|---|---|---|---|
| Each campaign | From Goals | From CRM source fields | Match rate | Sales stage | Upload report | CRM | Spend ÷ qualified |
Some gap is normal. Spam removed in the CRM, duplicate submissions and different dating of the same lead in each system all explain a few leads. A gap you can’t explain lead by lead is a rung 1 or rung 2 finding. The maths for the last column is in cost per qualified lead, and which stages to send back is in CRM lead stages for ad platforms.
Evidence sheet
Short answer
Rank measurement failures on rungs 1 and 2 first, whatever the spend, because every other finding depends on them. Rank the rest by spend affected multiplied by signal risk. Then change one thing per rung per cycle, so you can tell which fix moved qualified leads and which only moved the reported numbers.
| Column | What goes in it |
|---|---|
| Rung and check | Such as “1: WhatsApp click set as primary” |
| Evidence | Screenshot or export, with the date range |
| Spend affected | In ₹, AED or $ for the period |
| Signal risk | High, medium or low: does this corrupt what bidding learns from? |
| Owner | Marketing, sales, RevOps or agency |
| Next step and review date | Act, Review, Investigate or Wait, and when to look again |
How Zephra uses the four decisions is explained in how it works.
Worked numbers
A UAE engineering services firm spends AED 40,000 a month: 150 enquiries in its CRM at about AED 267 each, 45 qualified at about AED 889, 15 proposals and 3 contracts. Google Ads reports 190 conversions.
| Rung | Finding | Decision |
|---|---|---|
| 1 | 190 Google conversions against 150 CRM enquiries. The website form action counts every conversion, and a WhatsApp click is a primary action. | Act: count one per click; make the WhatsApp click secondary |
| 2 | 57 of 150 enquiries (38%) have no click ID or email match, because they arrived by WhatsApp or phone, so qualified imports undercount. | Review the capture route with sales |
| 3 and 4 | Performance Max search terms include brand queries, and brand exclusions are missing. | Investigate: brand exclusions and the channel performance report |
| 5 | Targets were set on the inflated count. | Wait one conversion cycle after the counting fix before changing targets |
| Campaign | Spend | Enquiries | Qualified | Qualified rate | AED per qualified |
|---|---|---|---|---|---|
| Search (services and brand) | AED 28,000 | 100 | 37 | 37% | about 757 |
| Performance Max | AED 12,000 | 50 | 8 | 16% | 1,500 |
| Total | AED 40,000 | 150 | 45 | 30% | about 889 |
Fixing the count will drop reported conversions from 190 towards 150. That changes the numbers, not the demand, so don’t read it as a performance fall.
Its 8 qualified leads may include brand demand Search would have caught. Judge it again once brand exclusions and a qualified-lead goal are in place.
A pre-launch check is the same ladder with nothing to read yet. Only these checks differ:
Cadence
Short answer
Review search terms weekly and reconcile leads with sales monthly. Run a full audit each quarter and after any major change, such as a new bid strategy, a changed conversion goal, an AI Max upgrade or a website form update. B2B accounts change slowly, so audit evidence rather than daily fluctuations.
| When | What |
|---|---|
| Weekly | Search terms, spam, change history |
| Monthly | Lead reconciliation and qualified-lead import health |
| Quarterly | The full ladder |
| After a major change | Rungs 0 to 2 |
Who audits
Short answer
Audit it yourself if someone can pull lead stages from your CRM or sheet and has a few hours for measurement and search terms. An outside assessment helps when the team or agency that built the account is judging its own work, or when nobody can connect Google’s numbers to sales outcomes.
Zephra’s free ad account assessment takes 10–15 minutes, reviews recent account history, and covers wasted spend, campaign structure, performance constraints and measurement gaps. It connects by OAuth, which you can revoke, needs no CRM and makes no campaign changes. Google, Meta and LinkedIn accounts can connect.
Context
| Size | Scope | Who owns what |
|---|---|---|
| SMB | Rungs 0 to 3 and 6 in an afternoon; reconciliation in a spreadsheet | Founder or marketing lead with the sales head |
| Mid-market | Full ladder quarterly | RevOps owns rung 2; an in-house lead reviews agency findings |
| Enterprise | A ladder per account across a manager account, starting with the highest-spend campaigns | Change control records for rung 0; privacy sign-off on data uploads |
WhatsApp and phone enquiries break click ID capture, which shows up on rung 2. IndiaMART and TradeIndia leads often get logged as “Google” in the CRM, so check source hygiene before reconciling; see B2B marketplaces vs paid ads. The default location option can pull in overseas job seekers. Give notice under the DPDP Act before uploading lead data.
Campaigns covering the UAE, Saudi Arabia and Qatar need reconciling per country. Google says Search campaigns prioritise ads matching the search term language from September 2026 (Google Ads Help), so review language settings. Check WhatsApp capture and your obligations under UAE data protection law.
CRM lifecycle stages usually become the qualified import. Check call tracking against consent rules for follow-up, and state privacy rules before uploading customer data.
Zephra monitors tracking and event health, spend anomalies and Performance Max health, and reads campaigns against lead quality from your CRM or a spreadsheet, where configured. See Google Ads for B2B and lead quality optimisation.
Questions, answered
Check, in order: recent changes in change history, whether the lead conversion still fires and counts once, whether campaigns are limited by budget or targeting the wrong locations, and whether search terms match what you sell. If leads arrive but sales rejects them, the problem is signal and intent, not volume.
For a small account, a few hours covers rungs 0 to 3. A mid-market account with several product lines usually needs one to two days, most of it spent pulling and matching CRM data.
Enough to see reports, change history, recommendations and conversion goals, plus an export of lead stages from your CRM or sheet. For an outside assessment, see the free ad account assessment page.
No. Google describes it as an estimate of how well an account is set to perform, calculated from settings, statistics and available recommendations. It says nothing about whether leads qualified.
Review which types are on, and keep only those you would accept every time without looking. Anything that changes matching, bidding or budget deserves a human decision in a B2B account.
Wasted spend, campaign structure, performance constraints and measurement gaps. It takes 10–15 minutes, looks back over 1 month, 6 months or 1 year, needs no CRM and makes no campaign changes.
Your account
The free ad account assessment takes 10–15 minutes and reviews wasted spend, campaign structure, performance constraints and measurement gaps. No CRM needed, no campaign changes during the assessment.