PUSHA PUSH product
A hand holding a phone. On the screen: your best ad has a secret.

MeetCreativeIQ

Know exactly why your ads work.

CreativeIQ opens every ad you’ve run, takes each one apart into 180+ data points with two agent swarms, finds the patterns that hold, tells you what to do next and writes the briefs for the next ads.

See it work Get a free teardown
  • Supplements
  • Outdoor clothing
  • Homeware
  • Garden buildings
  • Gold buying
  • Comparison
  • Fantasy football
  • Wellness
  • Schools
  • Travel insurance
  • Work boots
  • Footwear
  • Pet products
  • Door entry

Anyone can show you which ad won.Almost nobody can tell you why.

Ask “what’s working?” and the usual answer is the top five ads by returna gut feelinglast month’s winner, againwhoever argued loudest

CreativeIQ swaps the hunch for evidence.

Your ad report has a column for everything.

Spend, results, cost per result. Everything except the reason.

AdEffectWhy?
Garden buildings retailerOffer carouselDirectional
5x cheaperper website enquiry, against the brand’s feature film

The worst photograph in the account, carrying a finite-stock offer at a named site.

Comparison siteDesigned image
56% cheaperper lead, against the other test images

It argued the wait with a printed daily price. Fewer clicks, far cheaper leads.

Outdoor clothing brandProspecting films
19% cheaperper purchase, against films that never ask

They end by asking viewers to shop. Three of the four biggest films never asked.

Homeware retailerCrewed location film
41% dearerper purchase than the cheapest film to make

A quarter of the video budget went on the weakest film. An 18-second phone clip beat it.

Wellness clinicAdvice-led videoDirectional
82% fewer basketsper click, against films that end on the ask

The best hook in the account ended on advice, not an offer. Clicks with nowhere to go.

Gold buying serviceWhole account
+50% cost per leadwhile click-through rose 51%

The ads got better at earning clicks. The leak was the landing page, not the creative.

Meta, 2025 to 2026. Green is cheaper, red is dearer, each against the comparison named. Directional reads rest on a handful of ads.

How it works

One ad.

A creator video that brought new customers in 45% cheaper than the account’s average video. Why that one?

0 data points.

Every frame, word, sound and cut, pulled apart and logged.

Two agent swarms.

One swarm watches and measures. The other reasons about why each choice worked.

Finding 1 · Outdoor clothing

The biggest films never asked.

Every film in the account, sized by its spend. Films that end on an ask cost 19% less per purchase, and the three biggest never ask.

Finding 2 · Garden buildings

The worst photo won.

A plain snapshot carrying a finite-stock offer brought website enquiries in 5x cheaper than the brand’s polished film.

Finding 3 · Supplements

Show the brand by 0:02.

Every video in the account, sized by spend. Videos with the brand on screen inside two seconds won customers 35% cheaper. Take out the three biggest ads and the gap grows to 47%.

Three decisions.

What to do on Monday, with the ads behind each one.

01It watches

Two agent swarms

180+ data points an ad.

The first swarm watches: every frame, word, sound and cut, measured. The second reasons: why each choice worked and what it cost. Images are read blind to their results.

Readings from a real creator ad in a supplements account. The footage is a stand-in.

  • 0:00 · watchProduct in hand from the first frame
  • 0:05 · watchThe hook: an apology, not a pitch
  • Audio · watchOne take, no music, just her voice
  • ReasonTalking head, no music: the account’s cheapest format
  • ReasonThe proof is shown on her own hair and skin, not just said
  • ReasonShe wins twice: a creator effect, not a product one

45% cheaper per new customer than the account’s average video

02It finds the patterns

Pattern analysis

Every ad, side by side.

The same questions go to every ad. Each one splits the account in two and weighs the money on each side. Then it tries to knock the answer over.

  • Weighted by spend
  • Video and images kept apart
  • Stress-tested before you see it

Real ads from three accounts. Each dot is one ad, sized by its share of spend.

Brand readable in the first two secondsSupplements · video
35%cheaper per new customer

Left is cheaper than the account average, right is dearer. Ads that never converted sit at the far right.

03It decides and tells you

The Monday read

It tells you what to do next.

It reads every pattern that held, keeps the few that would change what you do, and writes them in plain English with the ads beside them.

  • Up to three decisions
  • Every number with its ads
  • One page, no dashboard
CreativeIQMonday read

Supplements brand

58%cheaper
Sell Magnesium through the review routeMagnesium only. The other products don’t move.
35%cheaper
Show the brand in the first two secondsHeld inside every product line
30%cheaper
Keep the offer printed on the imageHeld with the three biggest ads removed

Outdoor clothing brand

19%cheaper
End every prospecting film on an askHeld in three regions, 42% in the EU
35%cheaper
Put the offer on prospecting imagesGrows with the three biggest ads removed
8%dearer
Watch sale images in the UKThe one market where the offer reverses

Homeware retailer

6.9xcheaper
Send images to a collection, not one productHeld across three funnel stages
38%cheaper
Move film budget to mixed audiencesPer purchase, against mid-funnel
41%dearer
Retire the crewed location shootAgainst the cheapest film to make

Every decision arrives with the ads behind it.

The briefs

Then it writes the next ads.

Every decision becomes a brief: what it tests, why it should work, the first second, the film beat by beat and the rules. Built on the ads that proved it.

01 · Re-edit, no shootThe bestseller testimonial, now with an ask

Same footage and audience: does an ask alone bring the cost down?

02 · New filmThe waterproof, in the rain it is for

Does real rain plus a spoken ask beat the films that never ask?

Shown below
03 · New filmThe bestseller, straight out of the sea

In its own weather, with an ask: does it beat the original and the re-edit?

04 · New filmTrousers on the wet coast path

Does the ask-and-weather film hold for womenswear?

Brief 02 · New film · Outdoor clothing

The waterproof, in the rain it is for

23 seconds · 9:16 and 4:5
What it answers

Does a waterproof filmed in real rain, with a spoken ask and an end card, beat the films that never ask?

Where it runs

Prospecting, the UK and EU first, where the ask pays most.

Why it should work
  • 19%cheaper per purchase when films end on an ask
  • 42%cheaper in the EU, where the ask pays most
  • 1 in 5of the brand’s images show weather. Rivals: up to half
Shot 10 to 3s
Forecast said light showers.

Rain beading on the shell, walking up the ridge

Voice

“Forecast said light showers.”

Shot 23 to 8s

Hood up into the wind

Voice

“It lied. Hood up, and I’m still dry underneath.”

Shot 38 to 10s

No words. Two walkers climbing in fog, water running off a sleeve

No voice

Shot 410 to 16s

Close on the wearer, laughing in the rain

Voice

“This is what it’s for.”

Shot 516 to 20s
Shop the jacket

To camera, still in the rain

Voice

“Shop the jacket. Link below.”

Shot 620 to 23s
The waterproof jacketShop now

End card: the jacket’s name and the ask

No voice

Must
  • Real rain or driving wind on the shell in the first second
  • The wearer walking or climbing in two thirds of the shots
  • The ask spoken and on screen, then an end card
Must not
  • Bright or sunny footage
  • Sale badges, stars or bestseller badges
  • Technical claims the brand hasn’t supplied
Built from

The account’s own asking films, and the rivals’ longest-running weather films.

And nothing reaches you untested.

A pattern that only holds because of one big ad, one creator or one product is a coincidence. Every finding, decision and brief is stress-tested first, and anything that fails never reaches your page.

Comparison site · images

The problem argued with a printed daily price

56% cheaper per lead, against the other test images

  • Take out the three biggest adsStill 31% cheaper
  • Compare inside one ad set only53% cheaper
  • Look inside each age band5 of 5 agree
  • One placement only, the feed55% cheaper
HeldReached the client’s page
Fantasy football · video

Openers built on a testimonial

The strongest hook in the account, at first look

  • Take out the single biggest adHook rate falls 55%
  • Same test on structured openersFalls 45%
  • Same test on a casual toneFalls 38%
  • Music looked 16% cheaperGone without one creator
DroppedOne ad was carrying it

Six more looked obvious in one outdoor clothing account and failed: star ratings, customer quotes, subheads, studio shots, bad weather and bestseller badges.

Competitors

Your rivals, on the same ruler.

CreativeIQ pulls competitors’ live ads from the public Ad Library and asks them the same questions it asks yours. So “they do it and we don’t” gets counted, not eyeballed. Then your own account says whether copying them would pay.

No account access needed. It’s how every free teardown starts.

Videos that name the ingredient

A supplements brand against four rivals, UK
  • Rival A90%
  • Rival B75%
  • Rival C68%
  • Rival D63%
  • The brand55%
32% cheaper

In the brand’s own account, videos that name the ingredient bring a new customer in for 32% less, and it holds with the top three ads taken out. Copy the rivals.

Images that open on the problem

A comparison site against six rivals, UK
  • Rival A75%
  • The brand66%
  • Rival B14%
  • Rival C7%
  • Rivals D to F0%
56% cheaper

The brand’s strongest image argued the problem with a printed daily price, and it held under every test. The big insurers almost never open on the problem. Keep the lead.

Images that show cold, wet or windy weather

An outdoor clothing brand against four rivals
  • Rival A50%
  • Rival B50%
  • Rival C25%
  • Rival D25%
  • The brand19%
Don’t copy

Weather images looked 27% cheaper across the brand’s own account, then reversed in prospecting, where the money is. The rivals’ habit would not have paid.

Ads that ask for a visit or an enquiry

An independent school against three rivals
  • The school57%
  • Rival A11%
  • Rival B0%
  • Rival C0%
3.7x

Rivals build the brand and keep it running: their longest ads ran 3.7x longer than the school’s. The school asks hard and rotates fast.

Live ads from the public Meta Ad Library, coded the same way as the brand’s own.

The usual review looks at five ads.

3%of the ads watched
100%of the ads opened and read
97%never opened
96%of two years’ spend weighed
1hunch
180+data points on every ad

Most reviews stop at the top five. Everything else goes unwatched.

A gold buying account, two years of ads: every one that mattered, opened, read and weighed.

Sixteen brands. One set of rules.

Eight account runs and eight Ad Library teardowns, from supplements to work boots. The same questions and the same discipline every time.

Account run42%

cheaper per purchase in the EU when films end on an ask.

Outdoor clothing
Account run56%

cheaper per lead from one card that argued the problem with a price.

Comparison site
Account run6.9x

cheaper images when they land on a collection, not one product.

Homeware
Account run1.46x

the hook rate for medium-paced edits over fast ones.

Fantasy football
Account run5x

cheaper per website enquiry once the carousel carried an offer.

Garden buildings
Account run+51%

click-through while cost per lead rose. The leak was the landing page.

Gold buying
Account run5.5x

the baskets per click when the film ended on the ask.

Wellness clinic
Account run58%

cheaper per new customer selling one product through the review route.

Supplements
Ad Library teardown69%

of rival images lead with money off.

Travel insurance
Ad Library teardown6.8x

the median run for a rival’s ads against the school’s own.

Independent school
Ad Library teardown14x

longer in market for a rival’s boot film than the brand’s best.

Work boots
Ad Library teardown0%

of the brand’s catalogue cards carry its name. One rival brands every card.

Footwear
Ad Library teardown86%

of long-running videos show a person on the product by second three.

Pet products
Ad Library teardown0%

of the market leader’s images name a medical condition.

Travel insurance
Ad Library teardown1.7x

longer runs for quote ads than for the guide downloads.

Door entry

Start with your own ads.

Three ways in. The first one costs you nothing.

Free

Teardown

Your live ads and a competitor’s, read from the public Ad Library. Nothing to install and no access needed.

Get a free teardown
One-off

Look-back

Every ad in your account opened, counted and weighed against its spend. Up to three decisions, each with the ads behind it.

Ask about a look-back
Monthly

Every month

The same read on a schedule. It remembers what it learned on your account, so each run starts where the last one finished.

Ask about monthly

Not “what did well”. What to do next, and why we’re sure.

Get a free teardown
by PUSH
Free teardown

Get your free teardown

Your live ads and a rival’s, read from the public Ad Library. Nothing to install and no access needed.

What are you after?

We only use this to reply to you. Prefer to talk?