How to read a competitor's creative strategy from 30 active ads

A 3-phase framework for pulling 30 ads, finding the control, and spotting what they're testing this week.

Seeing ads isn't seeing strategy

Most operators open Meta Ad Library, look at 2 or 3 ads from a competitor, and close the tab feeling informed.

They saw the ads. They didn't see the strategy.

One ad tells you nothing, three ads tell you a vibe. Thirty ads tell you where the money is, what they trust, and what they're testing this week.

Every ad spy guide you've read is about finding ads. This one is about reading them.

I'll walk through the framework I use when I open Brandsearch Discovery, pull 30 active ads from a single competitor, and turn that into a picture of their testing cadence and budget bets.

Step 1: build the sample (the rule of 30)

Before interpretation, you need a sample big enough to trust.

Two ads are an anecdote, ten ads are a hint, thirty ads are a system. Below 30, you can't see the ratio of old to new, control to test, winner to throwaway.

Above 50, you start re-reading the same hook six times. Thirty is the sweet spot.

Here's how I pull the sample in Discovery:

1. Filter to one brand. Drop the competitor's domain in the brand filter.
2. Filter to one platform. Meta for now. Don't mix platforms, the testing logic is different on TikTok.
3. Filter by format. Video only if it's a video-first brand, mixed if you're not sure.
4. Sort by running days, descending. Oldest first.

You should end up with a grid where the top is the longest-running ads and the bottom is the newest. That ordering is the whole trick.

If the brand has fewer than 30 active ads, drop the platform filter and pull cross-channel. If they have 200, cap at 30 by diversity: take the 15 longest-running and the 15 newest.

Step 2: find the control (the cluster)

Now you read top-down. Oldest ads first.

An ad that's been running for 25+ days without pause is paying for itself. Meta cuts dead creative in the first week, so if a competitor keeps the same ad live for 40 days, the ad is working.

Here's the rule: count how many of your 30 ads share the same opening hook, the same offer framing, or the same visual setup. If 6 out of 30 open on a person pointing at camera saying "I tried this for 30 days", you've found the control cluster.

It's not one ad. It's a cluster. The cluster is the bet.

A real example. I pulled 30 active ads from a skincare brand in Discovery last month. Sorted by age. Top of the list: 8 ads, all 40+ days old, all opening on the same founder to-camera framing, same "my dermatologist told me" hook, same 15% offer.

That's 27% of the sample pointing at a single creative bet. That founder hook is their control.

Two things to look for inside the cluster:

• Duration: how long has the longest ad in the cluster been running? 25 days is my minimum threshold for calling something a control.
• Variants: how many slight variations exist? 8 ads using the same hook with 4 different voiceovers means they're scaling it, not testing it.

Spotting the control is 80% of the analysis. Once you know what's carrying their revenue, everything else in the sample makes sense.

Step 3: spot the test (the delta)

Now you read the same 30 ads bottom-up.

You're looking for ads that entered the rotation in the last 7 to 14 days. These are the tests.

Three things to read off the tests.

Hook variance. Is the new cluster using a completely different opening?

If the control is a founder to-camera and the new batch is UGC unboxing, they're testing a full angle swap. They think the current control has a ceiling.

Volume. One new ad is a check, but six new ads with similar framing is a campaign. The volume tells you how committed the brand is to the new angle.

Consistency with the old wins. Sometimes the new test is a direct descendant of the control with the same hook, a fresh actor, and a new offer, which is iteration, not a pivot.

In the skincare example, the bottom of the list had 5 ads from the last 10 days, all opening on a side-by-side transformation instead of the founder pitch. That's a pivot test.

You just learned three things about that brand's next 30 days without talking to anyone:

1. The founder to-camera angle is starting to fatigue.
2. They're betting on before/after as the replacement angle.
3. They're willing to spend on 5+ variants to find out.

That's strategy. That's what 30 ads tells you that 3 ads can't.

Cross-check with the Brand Analysis page

The 30-ad pull tells you what's happening right now. The Brand Analysis page tells you whether that behavior is normal.

Open the brand in Brand Analysis. The Overview tab shows the ad scaling chart and traffic trends, and that chart tells you whether the current 30 ads is a peak, a trough, or a normal Tuesday.

A brand running 30 ads during a usual 80-ad month is pulling back, while a brand running 30 ads during a usual 12-ad month is about to scale. Same 30 ads, opposite signal.

I also check the Scripts tab when the brand is video-heavy. Scripts gives you the transcribed opening lines for every video ad, and if your cluster analysis said "6 of 30 ads open with the same hook," it tells you whether it's literally the same copy or just the same angle performed differently.

Same copy means scaling. Same angle means variant testing.

Pattern-matching across the niche

Reading one brand is tactical. Reading 5 brands in the same niche is where this gets dangerous.

Pull 30 ads from each of 5 competitors using Brandsearch Discovery and Brandsearch Brand Library. Same filters, same sort, same process.

Now you're looking at whether every brand in the niche is running the same control.

When 4 out of 5 brands all run the same hook as their control, you've found the niche meta, the angle the whole category has validated. Your move: find out what the next angle is going to be, because the meta is already priced in.

When 4 out of 5 brands are all testing the same new hook in the last 14 days, you've found the early signal on a shift. You're watching 4 different marketing teams place the same bet independently, and that's worth more than any trend report.

From spy to strategist

Here's the full framework, stripped down:

1. Pull 30 ads from one competitor: Brandsearch Discovery, one brand, one platform, sort by age.
2. Find the control: read top-down. The cluster that repeats 5+ times is the money ad. Write down hook, offer, visual format.
3. Spot the test: read bottom-up. New ads from the last 7-14 days are where the fresh budget is going.
4. Cross-check: Brandsearch Brand Analysis Overview ad scaling chart. Is the current ad count normal or anomalous?
5. Pattern-match: Brandsearch Brand Library. Run the same 30-ad read on 5 brands in the niche to find the meta.

If you want a free entry point, start with the Brandsearch Chrome Extension. It pulls up traffic, ad count, and tech stack on any Shopify store you land on, and when a brand looks worth tracking you can carry them into the full app and run the 30-ad method.

Pull 30 ads this afternoon from one competitor, sort by age, and read top-down for the control. Read bottom-up for the test and write the two clusters down on a post-it.

That post-it is worth more than every "winning creative framework" PDF on the internet.