How I test AI ad generators

Ten tests, the same for every generator, each graded from 1 to 5 on a fixed scale. The overall grade is their average plus my verdict, which can move it by up to one point. Every test is run by hand on an account I use, and the ads, exports, counts, recordings and screenshots behind each grade go in the full test results published with the review.

Written by Miles Foster, Owner and tester, RankDude.

The RankDude mascot holding a laptop with a check on its screen

Each generator gets the same three products from fictional brands: a skincare serum, a running shoe and a project management app. Each comes with the same brand kit (a logo, two colours as hex codes, one font), the same product photos, a product page, a fact sheet of eight true claims and a list of claims that must not appear. I ask every tool for ten static ads and three short videos per product, and push finished ads to test ad accounts. Nothing is graded from a showcase or a press kit.

Each review states how the account was obtained: bought at the public price, used on a free tier, or provided by the vendor. A vendor-provided account never changes a grade, and a vendor never sees a review before it is published.

How I rate

Every tool gets one score out of 5.

First, I run the same 10 tests on every tool and give each one a grade from 1 to 5. The average of the graded tests is the starting point; a test that cannot be graded is left out of it, and the review says how many tests the score is based on.

Then I add my verdict: what the tests cannot measure, such as how the tool holds up after the test weeks, how billing and cancelling work in practice, and how it compares with the others in the category. My verdict can move the score up or down by one point at most.

Every review shows how much my verdict changed the score, and why, in its scores table.

What a grade means

5	Excellent
4	Good
3	Average
2	Below average
1	Poor

The final score is the average of the graded tests (ten when every test can be graded), to one decimal place, plus my verdict, which can move it by up to one point. If a test couldn't be finished properly it gets no grade, and the review says how many tests the score is based on. Some tests on a free plan use fewer ads or videos than usual because of credit limits; I test them again on a paid plan when I can. A tool that makes only video, only copy or only product photos gets no grade on the tests it cannot attempt, and the review says which.

The 10 tests

1.Creative Quality

Each of the 30 static ads is viewed at phone size and checked for six points: the product shown accurately (shape, label, colour), text legible, a clear visual order, a visible call to action, no image artefacts, and not taken for AI-made by a panel of three people. An ad passes with at least five of the six.

5: 27 to 30 pass
4: 22 to 26
3: 15 to 21
2: 8 to 14
1: under 8

2.Brand Fit

I save the brand kit in the tool once, where it allows, and generate without re-entering it. Each ad is checked for the logo unaltered, colours within 3 delta E of the hex codes (sampled), the brand font, and the tone set in the brief. An ad counts as on-brand only when all four are right.

5: 28 to 30 on-brand
4: 24 to 27
3: 18 to 23
2: 10 to 17
1: under 10, or no brand kit

3.Copy & Claims Accuracy

Every claim across the 30 static ads and 9 video scripts is checked against the fact sheet. Invented claims, claims on the banned list (medical promises, unprovable superlatives, fake discounts), wrong prices and spelling or grammar errors are counted separately.

5: no invented or banned claims, no errors
4: one minor error
3: one invented claim, or two or three errors
2: any banned claim, two or more invented, or four to six errors
1: invented or banned claims throughout, or seven or more errors

4.Variety of Variants

From the first 20 variants for each product I count distinct concepts: a different hook, angle or layout. A colour swap, a reworded headline or the same layout with a new photo counts once. The average across the three products is the grade.

5: 15 or more on average
4: 11 to 14
3: 7 to 10
2: 3 to 6
1: 2 or fewer

5.Platform Sizes & Formats

For one ad I request 1:1, 4:5, 9:16, 1.91:1 and 16:9, plus 300x250, 728x90 and 160x600 display banners. Each file is measured in pixels, checked for a sensible re-layout rather than a crop that cuts off text or product, and the 9:16 version is checked against the platforms' published safe zones.

5: all eight correct, safe zones respected
4: seven
3: five or six
2: three or four
1: two or fewer

6.Video & Avatar Quality

Each product gets three videos of 15 to 30 seconds, with an AI presenter where offered. Lip sync is measured frame by frame at 30 frames per second, and each video is checked for face and hand artefacts, the product shown correctly, captions matching the voice, and a natural voice. The worst video sets the floor.

5: lip sync within 2 frames, no artefacts, product right in all nine
4: minor artefacts in one or two
3: artefacts in three to five
2: product wrong, or lip sync off by more than 5 frames
1: no usable video

7.Speed & Ease of Use

For each product the clock runs from pasting the product page or brief to ten static ads ready to export, and separately to one finished video, with my fixes included. Steps are counted and dead ends noted. The median across the three products is the grade.

5: ten ads under 10 minutes and a video under 20
4: under 20 and 40 minutes
3: under 40 and 60 minutes
2: longer
1: could not finish

8.Ad Platform Integration & Testing

Each point is tried on test ad accounts on Meta, Google and TikTok: download images as PNG and JPG at full size; download video as MP4; push to each of the three as a draft with copy landing in the right fields (three points); set up an A/B test of two variants as a draft; connect each account's reporting so its campaigns and result fields show in the tool; and update an ad already pushed as a draft from inside the tool after an edit. Nothing is spent, so the reporting point checks the connection, not the results.

5: all eight
4: six or seven
3: four or five
2: two or three, downloads only
1: no clean export

9.Rights to Use

I read the terms of the plan used, and the licences named for its assets, for five points: stock images licensed for paid ads, presenter likenesses used with consent and cleared for commercial use, music cleared for paid social, outputs owned by me, and no plan limit on paid use. A point counts only when the terms state it.

5: all five in writing
4: four
3: three
2: one or two
1: unclear, or paid use forbidden

10.Pricing & Value

Every plan and credit pack is read from the checkout, not the marketing page. The cost of one finished static ad and one 30-second video is worked out in dollars on the cheapest plan with no watermark. Credit rules for retries, free-tier limits and payment methods are listed as seen at checkout.

5: under $0.50 per static and under $5 per video
4: under $1 and $10
3: under $2 and $20
2: more, or unclear credit costs
1: poor value or hidden costs

Documented but not graded

Competitor ad research, landing page builders, team approval flows, translation into other languages and creative performance predictions are checked and described in every review. When they work well they count in the written assessment, but they never add points to the overall grade.

Rules that apply to every test

Real use of the product on a real account, never promotional descriptions or showcase galleries.
Identical products, brand kits, tests, thresholds and sample sizes for every generator in the category.
Authorised accounts only; no password sharing, no bypassing of credit limits, watermarks or bot checks, and test ad accounts that never spend on live campaigns.
Every grade has evidence: the ads and videos, a count, a measurement, a screenshot or the terms quoted, in the full test results.
A grade changes only when a retest produces new evidence; the review records the date and the reason.

Questions about a specific grade go to the contact page. Makers can submit an AI ad generator for testing. No AI ad generator has been through these tests yet; every one I follow is listed on the AI ad generators page, and reviews appear there as tests finish.