Competitor ad audit: a six-step template with real benchmarks

Search "competitor ad audit" and most of what comes back audits your own account — a PPC health check on wasted spend and Quality Score. That's a different job. This is the other one: a structured review of what your rivals are actually running, in a shape you can put in front of a client. We've written it as six steps, and at every step we've attached a measured benchmark from the 50,831 ads we've qualified across 158 UK brands, so each line of your audit says "20% off is soft for this channel" instead of "20% off seems low". Figures refreshed 3 November 2026.

Why the usual competitor ad audit falls over

Two kinds of content dominate this query, and neither does the job.

The first is PPC account auditing — search term reports, negative keywords, conversion tracking hygiene. Genuinely valuable work, but it audits your account, not your competition. If you arrived with a client asking "what are the other four firms in our category doing?", it cannot answer that.

The second is the honest competitor-framed template. The best one we've read is Eden's 30-minute competitor ad audit, and its structure is sound — scale, proven layer, testing layer, gaps, one-page output. But every number in it is illustrative by its own admission: "~40 active ads", "70% UGC video" are examples of what you might write, not what normal looks like. That's the weak point of every template here. A template tells you which boxes to fill, not whether the value you wrote in the box is high or low.

That distinction decides whether an audit survives a client meeting. "Competitor A is running 20% off" invites the obvious question — is that a lot? A shrug makes the audit a screenshot collection. "20% is below the 38.1% average discount depth we see on Meta, so they're protecting margin rather than buying share" is something they can act on.

Step 1: Set the scope — three to five rivals, not fifteen

Pick three to five competitors and do them properly. This is the one piece of received wisdom in this space we fully endorse, because the failure mode of a wide audit is predictable: you collect fifteen brands' thumbnails, run out of attention at brand four, and present a table with half-empty rows.

Choose on a defensible basis and write the basis down in the audit. Usually it's one of: the brands bidding on your client's brand name (see competitor brand bidding), the brands appearing in the same category SERPs, or the brands the client's sales team keeps losing deals to. All three are legitimate. "The ones we thought of" is not.

Benchmark: what a normal ad count looks like

Volume is the first thing a client asks about and the easiest to get wrong, because perceptions are anchored on the biggest advertisers. Across the brands we track, qualified live ad counts spread very wide — the largest in our set sits near 1,957 qualified creatives (Pets at Home), with ASOS at 1,752 and easyJet at 1,459, while plenty of serious mid-market advertisers run in the low hundreds. So if your rival has 40 live ads, that is neither impressive nor alarming on its own — it's a mid-market footprint. We go into the distribution in detail in how many ads do competitors run.

One caveat to record in the audit, because it will otherwise undercut your own numbers later: Google's Transparency Center stops paginating at roughly 900 creatives per advertiser. That isn't a rare edge case: 48 of the brands we track hold more than 900 Google creatives, so for those advertisers any count you or anyone else publishes is a floor, not a total.

Step 2: Map the channel mix before judging the creative

Before you look at a single hook, count where the ads are. Channel mix is the cheapest strategic read in the audit, and the thing clients most often get wrong about their own market, because they only watch the channel they personally buy.

Platform splits are genuinely diagnostic. In our data Pets at Home runs 964 qualified creatives on Google, 794 on TikTok and 199 on Meta — a brand that has committed real creative resource to short-form video. ASOS, by contrast, sits at 1,506 on Google against 165 on Meta and 81 on TikTok. Same retail category, very different posture. Neither is wrong, but if your client's plan is "more Meta because that's what everyone does", that mix is the slide that kills the assumption.

Benchmark: creative conventions differ sharply by platform

Judge creative against its own channel's norms, not a single global standard. Human faces appear in 51.7% of the TikTok ads we've qualified and 44.6% of Meta ads, but only 2.8% of Google ads — Google's inventory is dominated by product and text formats where a face would be unusual. A client's "our competitors all use people in their ads" is usually a statement about which platform they've been looking at.

Step 3: Score the offer — and check the depth against the channel

This is where the audit earns its fee, and where the benchmark matters most. For each rival, record the offer type, the discount the ad actually states, whether a promo code is visible, and any deadline or condition. Four fields, every ad. Where the ad gives no figure, record "none stated" rather than guessing at the brand's usual offer — that distinction is what keeps your depth averages honest.

Then compare against the channel. Offer prevalence and depth are not constants:

Read across that table and the same competitor ad means opposite things depending on where it ran. A 20%-off Meta ad is distinctly soft against the 38.1% Meta average — this rival is holding margin while others buy volume. The same 20% on TikTok sits above the 25.2% norm, in a channel where only 3.4% of ads carry any offer at all, making it an unusually promotional play. One number, two opposite conclusions — and that is the sentence a template without data cannot write for you.

Promo codes deserve a note of their own because their rarity is the finding. Visible voucher codes appear in just 2.2% of Google ads, 1.3% of Meta ads and 0.2% of TikTok ads. So a competitor putting codes in paid creative is doing something deliberate — usually running affiliate or email crossover through paid, which means there's a channel in their mix your client may not have mapped. The brands that do it stand out hard: LookFantastic accounts for 276 code-carrying creatives in our set, with Cult Beauty next at 191 — beauty retail is where the tactic concentrates. If you want the mechanics of watching this over time, we cover it in tracking competitor promotions.

Step 4: Classify the hook, not just the headline

Clients can read a headline themselves. What they're paying you for is the pattern across a hundred of them. Sort each ad into a hook category and count. We use nine: bold claim, discount-led, problem-solution, aspirational, social proof, curiosity, question, stat, testimonial.

Benchmark: the hook distribution across 50,831 ads

Two findings here are directly usable in a gap analysis. First, bold claim and discount-led together take 46.4% of all ads — nearly half the market is shouting either "we're the best" or "we're cheap". Second, the proof-based hooks are close to vacant: social proof, stat and testimonial sum to 10.7%. When your client asks what nobody in their niche is doing, that's frequently the honest answer, and it's cheap to test. Our hook examples guide has the worked wording.

Step 5: Read the CTA mix as a funnel diagnosis

The call to action is the most under-audited field in competitor work, and the most revealing: it tells you which stage of the funnel a rival is buying. Count them:

Compare a single rival's mix against that baseline and the conclusion writes itself. A brand at 90%+ "shop now" is running a pure direct-response account with no upper funnel — ASOS sits at 93% and Victorian Plumbing at 98% in our data, which is the correct posture for high-intent retail but leaves awareness entirely to someone else. A brand heavy on "learn more" and "sign up" is buying consideration, usually because the purchase needs research. And a brand with a large "none" share is often running brand or video inventory where no button exists.

The useful audit line is the mismatch: a rival whose CTA mix is 80% "shop now" while their landing pages are long-form educational content has a funnel seam you can exploit. Checking that requires actually opening the destinations — see finding competitor landing pages.

Step 6: Separate proven ads from current bets

The final analytical step, and the one that converts an audit into a test plan. Split each rival's ads into two groups: long-running creatives (their proven playbook — nobody keeps paying for a loser) and recently launched ones (their current bets). The gap between the two groups is their direction of travel.

One hard caveat, because we've made this mistake ourselves in client-facing work. Ad libraries expose a duration figure, and it is tempting to read a high number as "recent activity". It isn't — duration tells you how long an ad has run, not whether it launched recently. A creative running 400 days is maximally proven and minimally new. The testing layer needs the first-seen date, and that is only reliable for ads that appeared after you started watching. We set out what launch dates do and don't support in how accurate is ad launch data.

This is the structural limit of any single-sitting audit: a one-off snapshot gives you a strong proven layer and a weak testing layer. The newest-ads section of a first audit is partly guesswork, and it's more honest to say so in the document than to present it with the same confidence as the rest. The testing layer only becomes trustworthy from your second capture onwards, which is the real argument for running the audit monthly rather than once.

The one-page output

Everything above is working. What the client sees should be one page, and every line should carry its benchmark:

1. Scale and mix — ad counts and platform split per rival, with the note that Google counts above ~900 are floors.
2. Offer posture — each rival's discount depth against the channel average (38.1% Meta, 30.0% Google, 25.2% TikTok), flagging who is buying share and who is protecting margin.
3. Message pattern — the hook distribution, against the 46.4% bold-claim-plus-discount baseline.
4. Funnel posture — CTA mix against the 43.3% shop-now baseline.
5. The gap — one named opening, ideally evidenced. Proof-led hooks at 10.7% of the market is the most common defensible one.
6. One test — a single prioritised recommendation, with the benchmark that justifies it.

How often to repeat it

Monthly per core competitor is a sensible floor, with one exception: promotional periods need weekly checks, because discount cadence moves in days and a sale spotted three weeks late is history, not intelligence. Our Black Friday tracking guide covers that compressed cadence, and competitor ad alerts sets out the full cadence table — daily, weekly, monthly, quarterly — along with which observations are worth reacting to and which are noise.

If you want spend context alongside the creative read, estimating competitor ad spend in the UK sets out which inferences the public data genuinely supports and which are fabrication. And if you want the full field-by-field rubric behind the benchmarks on this page rather than the audit workflow, that's competitor ad analysis.