Where should an ad account audit start?

With conversion tracking, not with bids. If the wrong event is counted, or counted twice, every later conclusion rests on wrong numbers. Check what is recorded as a conversion and whether the count matches your CRM.

What can I check if I only have half an hour?

Four things in order: which events count as conversions, where spend went with no conversion at all, where campaigns compete for the same query, and which partner networks and extra placements are switched on by default.

Should I change anything during the audit itself?

No. An audit produces a list with a reason and a data volume for each line. Changes come afterwards, one at a time, so the effect of each one can still be told apart from the others.

A 30 minute ad account audit: what to check, in order

The order to check an ad account in: conversion tracking, spend without results, internal competition, and default placements.

ADS Beast editorial teamPublished September 3, 2026Updated September 14, 20263 min read

Half an hour is enough to decide whether the numbers in an account can be trusted. Order matters more than coverage: start with the things that invalidate every other conclusion, not with whatever catches the eye first.

In short

Conversion tracking first, everything else after.
Read spend without results line by line, not as a total.
The same query in two campaigns means bidding against yourself.
Partner networks and extra placements are on by default almost everywhere.
An audit ends with a list, not with changes.

1. What counts as a conversion

Open the conversion actions and look at which ones are marked primary. A common picture: a form submission, a click on the phone number, and a visit to the contact page are all primary. The account then reports plenty of conversions while the sales team sees the same number of enquiries as before.

The second thing visible immediately is double counting. The same event arrives from the browser pixel and from the server side, and without a shared identifier it lands twice. One day reconciled against the CRM settles it.

Аудит рекламного аккаунта за 30 минут. Конверсии: Проверьте primary-действия и двойной учёт событий; Расход без результатов: Строки с тратами и нулём конверсий, отдельно поисковые запросы; Конкуренция внутри: Повторяющиеся запросы в разных кампаниях — аукцион против себя; Площадки: Партнёрские сети — Порядок проверок: от конверсий к настройкам, чтобы цифры можно было доверять

2. Where the money went with nothing to show

Build a report by campaign, ad group and search term for the last 30 days and sort by spend. Look at rows with spend and zero conversions. Read the search terms separately: that is usually where you find part of the budget going to a subject that is not yours.

Zero conversions is not a verdict on its own. What matters is how much the row has spent: once it is well past your usual cost per lead, luck stops being the explanation.

3. Competition inside the account

The same query enabled in two campaigns turns the account into an auction against itself. The tell is a query that wins impressions in one campaign, then the other, while the cost per click rises without a better position.

Check it with a list: export active keywords with their campaign and look for repeats. This is common in accounts that several people have run in turn.

4. Where the ads actually appear

Extra networks are on by default nearly everywhere: Audience Network on Meta, Search Partners on Google, Pangle on TikTok. Their quality differs from the main feed and from search, and the standard report adds all of it into one line.

Break spend and conversions down by placement. If the extra networks take a visible share of spend and return almost nothing, you have found the fastest way to move money back into the part that works.

5. Settings that change what the numbers mean

Three places where the platform counts more broadly than you expect: the attribution window, automatic audience expansion, and placement on partner sites. None of them is a mistake, and each of them changes what the word conversion means in your report.

What to do next

Collect the findings as a list: the line, the reason, and how much data stands behind it. Then change one item at a time and wait for a comparable period. Ten edits in one afternoon leave you no way to tell which one worked.

All blog articles