Paid Search Audit: What Your Campaigns Pay For That SEO Could Capture

A paid search audit checks that a Google Ads account spends well, but it does not stop at CPCs. It must show what your campaigns pay for that SEO could capture, and what SEA (paid search) covers that organic will not take. The stakes are real: 38% of the digital marketing budget is allocated to SEA (Odiens, 2025). This method cross-references Google Ads, Google Analytics 4 and Google Search Console, and extends the SEO vs SEA comparison with a decision grid.

What a paid search audit must decide

A standard audit measures efficiency: does the account spend in the right place, on qualified queries, with reliable measurement? That reading remains necessary, but it leaves out the question a B2B marketing manager actually asks: what does SEA pay for that SEO would deliver, and the other way round?

To answer it, the audit breaks the account ROAS down into four ROAS, calculated from Google Ads, Search Console and GA4:

• Overall ROAS: conversion value divided by total spend, the figure shown in the interface.
• Non-branded ROAS: the same calculation limited to generic queries.
• Branded ROAS with cannibalisation: brand campaigns, with all their conversions.
• Branded ROAS without cannibalisation: the same campaigns, once the conversions SEO would have captured are removed.

The gap between the last two measures what you pay for traffic you already had. The gap between overall and non-branded shows how much the brand flatters the average. Each of the following sections feeds one of these calculations.

Preparing the data: Google Ads, GA4, Search Console

An audit is only as good as its data. Before reading anything, check that the three sources describe the same conversions over the same period.

Measurement prerequisites

These checks are done in Google Analytics 4, since Universal Analytics shut down in 2023:

• Conversions imported from GA4: only useful key events (demo, qualified form, call) are sent to Google Ads.
• De-duplication: a conversion is not counted by both the Google Ads tag and the GA4 import.
• Offline import: in B2B, opportunities come back from the CRM, otherwise your ROAS remains a form ROAS.
• Data-driven attribution: it has been the default model since first-click was retired; compare periods read under the same model.
• Analysis window: 4 to 12 weeks, with a comparable reference period if the business is seasonal.

On the organic side, our Google Analytics guide helps you configure key events, and our Google Search Console guide helps you isolate branded queries.

The indicators to gather

Each indicator has a reference source and an associated decision:

These indicators only make sense once compared with commercial objectives: the article on SEO KPIs covers the organic side of this dashboard.

Segmenting: branded, non-branded, campaign type

An account average is almost always misleading. A comfortable ROAS can hide loss-making generic campaigns, offset by branded queries that would have converted anyway. Segment along at least four axes:

• Branded / non-branded, including branded queries that slip into generic campaigns.
• Campaign type: Search, Performance Max and Demand Gen have neither the same intent nor the same transparency.
• Device: mobile, high in volume but low in opportunities, drags CPA up.
• Region: regional differences will also serve the incrementality test.

Performance Max deserves particular attention. These campaigns mix networks and queries, can capture branded queries, and their search terms report remains partial. First ask what share of their conversions comes from your brand: it is the first blind spot of an audit.

Queries and budget: where money goes with no return

The Search terms report shows what users actually typed before clicking: this is where most of the waste shows up.

Search terms, match types and negative keywords

Sort search terms by cost, then check:

• Match types: broad widens delivery considerably, phrase and exact narrow it.
• Negative keywords: a shared list (jobs, internship, free, training), supplemented per campaign and justified.
• Branded queries in generic campaigns, which inflate non-branded ROAS.
• Intent: an informational query rarely converts and is better served by organic content.

The last point links the audit to SEO: only queries with high commercial value deserve a bid, and the article on high-intent keywords helps you tell them apart.

Lost impression share and click fraud

Lost impression share is read in two columns. Lost to budget, it signals profitable demand that is not funded. Lost to rank, it signals a quality or bidding problem that more budget will not solve.

Budget can also go on worthless clicks. The estimated click fraud rate in SEA reaches 11.7% (Odiens, 2025). Compare click spikes with engaged sessions in GA4 and with CRM contacts: a lasting gap on a region or time slot justifies an exclusion. More benchmarks are in our SEA statistics.

Quality Score, ads and landing pages

Quality Score is assigned by Google to each keyword. It influences CPC and impression share lost to rank. Google breaks it down into three components:

• Expected click-through rate on the query.
• Ad relevance to the query.
• Landing page experience, including on mobile.

The most profitable check is still manual. Take the 10 to 30 most expensive queries and follow the full chain: query typed, ad shown, page reached, conversion triggered. A promise missing from the top of the page, or a generic page for a precise query, explains many high CPAs that are wrongly blamed on bidding.

A page that converts paid traffic poorly converts organic traffic poorly too: the levers are covered in the article on conversion rate optimisation.

Smart Bidding: check the signal before the algorithm

The audit does not judge automation itself: it checks that the algorithm learns from the right signal. A target CPA fed with unqualified forms optimises for unqualified forms. Check:

• The strategy: target CPA, target ROAS or maximise conversions, depending on whether conversion value is known.
• Signal quality: useful contacts, de-duplicated, enriched by offline import.
• Volume: a rare conversion, such as a signed contract, gives too few examples; a qualified intermediate step helps the algorithm.
• The learning phase: every major change restarts it, so document changes and do not modify everything at once.

Brand cannibalisation and incrementality

A brand campaign almost always shows the best ROAS in the account, because it reaches users who are already searching for your company. The real question is how many of those clicks would have arrived without an ad, through the first organic result and its sitelinks.

Cross-referencing Search Console and Google Ads

Brand cannibalisation is measured by comparing three pieces of information, branded query by branded query:

• in Search Console, organic impressions, clicks and position;
• in Google Ads, paid clicks, CPC and conversions from the Search terms report;
• in Auction insights, whether or not competitors are present.

A query where the site holds organic position 1, with its sitelinks and no competitor, is the typical case of cannibalisation: the ROAS looks flattering, but a large share of the value would have been obtained without paying. How clicks split between the ad and the organic result is covered in the article on SEA vs SEO click-through rate.

Testing incrementality

Cross-referencing gives a hypothesis; only an incrementality test gives a measurement. Pause brand campaigns in one region or for one period, then compare total clicks and conversions, paid and organic, with a control region or period. If the total drops only slightly, branded paid search was mostly cannibalisation. If it drops sharply, it brought traffic that organic does not make up.

The following table links common situations to the signal and the decision:

Keeping branded paid search is therefore not a mistake in itself: it is a choice to justify with a competitor on your brand or an offer to push. The overall logic is developed in the article on the relationship between SEO and SEA.

Reading the competition without a third-party tool

Google provides the signals you need to read the pressure on your queries, without a competitive intelligence tool. What stays invisible are a competitor's budgets, bids and targeting.

Auction insights: what they show and their limits

The Auction insights report compares your presence with that of the advertisers in the same auctions, using four metrics:

• Impression share: yours and each competitor's.
• Overlap rate: how often their ad appears at the same time as yours.
• Position above rate: how often their ad appears above yours.
• Outranking share: how often yours appears above theirs, or alone.

These metrics are read by campaign, ad group or keyword, and say nothing about competitors' budgets or profitability. A rise in overlap signals pressure, not a strategy.

Google Ads Transparency Center

The Google Ads Transparency Center shows the ads an advertiser runs, by region and by format. It lets you check whether a competitor bids on your brand, with which messages and towards which pages. Cross-referenced with Auction insights, it helps you spot bidding wars, where your cost rises without your conversions following.

Three outcomes for each query

A gap analysis of paid queries places each query in one of three outcomes: win in SEA, build in SEO, or ignore.

Ads and AI surfaces

At the time of writing, AI Overviews and AI Mode have been available in France since summer 2026, and Google also shows ads in them. Some users are satisfied with the generated answer.

This shift weighs on organic clicks. The click-through rate of the first organic position drops to 2.6% when an AI Overview appears (Squid Impact, 2025). A top organic position therefore no longer guarantees the same volume of clicks, and your reading of cannibalisation must take this into account.

The audit draws three consequences:

• Record: note whether an AI Overview appears on your branded queries and on the most expensive generic ones.
• Incrementality: on these queries, an ad may bring more incremental clicks than before; rerun the pause test.
• Zero-click visibility: being cited in the generated answer is a matter of GEO, not bidding; track it outside ROAS.

How ads work on Google is explained in the article on paid search on Google, and visibility in generated answers in our guide to Generative Engine Optimization.

From findings to decisions: audit grid and action plan

A useful audit ends with decisions, not a list of findings.

The audit grid

For each checkpoint, the grid gives the SEA action, the SEO action and the priority, to adapt to your account:

The queries to transfer to SEO are, in order of priority:

• non-branded informational queries, which cost clicks without bringing contacts;
• queries where the site is already in organic position 1 to 3, with no competitor pushing;
• stable queries of known value, which an organic page can capture durably.

30- and 90-day plan

The plan does not promise results: it orders the tests and sets a decision criterion for each.

• 30 days: fix measurement, isolate the brand, add the obvious negative keywords, calculate the four ROAS, launch the pause test in one region.
• 90 days: read the test, reallocate the freed budget to profitable non-branded queries, strengthen the organic pages for the queries to transfer.
• Then: repeat the audit every quarter, and sooner after a change in bidding, site or offer.

To put a value on the transferred queries, the article on SEO ROI details the calculation on the organic side.

Depending on your situation

What comes next depends on what the audit revealed:

• if organic pages are missing, start with an SEO audit of the site;
• if paid queries mix intents, the article on search intent helps you classify them;
• if brand cannibalisation dominates, give priority to the pause test.

Calculating the four ROAS with Incremys

Incremys is a GEO & SEO platform: it does not manage campaigns and does not sell SEA audits. Its SEO vs SEA module automatically calculates the four ROAS (overall, non-branded, branded with and without cannibalisation) from Google Ads, Search Console and GA4. The opportunity analysis module spots paid queries that SEO can take over, and the 360 SEO & GEO audit checks that the target pages are ready.

Paid search audit FAQ

What is a paid search audit?

It is a diagnosis of a Google Ads account that checks spending efficiency (queries, ads, bidding, measurement) and what the campaigns pay for that SEO would deliver. It cross-references Google Ads, GA4 and Search Console, and ends with a decision for each spending segment: keep it, reduce it or transfer it to organic.

How often should you audit your campaigns?

Once a quarter. This pace gives Smart Bidding time to stabilise its learning and covers an analysis window of 4 to 12 weeks. Bring the audit forward after a major change: bidding strategy, site redesign, new offer, or a competitor appearing on your brand.

How do you measure brand cannibalisation?

Cross-reference, query by query, Search Console (organic clicks and position on the brand) and Google Ads (paid clicks, CPC, conversions). A brand in organic position 1, with no competitor bidding, signals likely cannibalisation. Only a pause test by region or by period truly measures it.

Should you switch off brand campaigns?

It depends on the incrementality test. If total paid and organic conversions remain stable during the pause, branded paid search was mostly cannibalisation. Keep it, on the other hand, if competitors bid on your brand, or to push an offer that the organic result does not show.

What should you read in Auction insights?

Four metrics: impression share, overlap rate, position above rate and outranking share. They show who takes part in the same auctions as you and how often your ads meet. They say nothing about competitors' budgets or profitability.

When can SEO replace a campaign?

When the query is non-branded and informational, when the site is already in organic position 1 to 3 with no competitive pressure, or when an organic page can durably capture stable demand. The transfer is made gradually, while tracking total paid and organic conversions.