RankDude

How I test AI design tools

Nine tests, the same for every tool, 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, the designs are made for real and exported for real, and the timings, screenshots and checklists behind each grade go in the full test results published with the review.

Written by Miles Foster, Owner and tester, RankDude.

Each tool gets the same three written briefs: a bakery's social media set (an Instagram post, a story and a Facebook banner), a five-screen booking app for a photographer, and a one-page landing site for a small shop with its own logo, two fonts and five colours. Each brief lists 20 required elements: the exact text, the sizes, the colours, the images supplied and the layout rules. I generate the first draft, then edit, export and share what the tool made. Nothing is graded from a feature list, a showcase gallery or a launch video.

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 9 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

5Excellent
4Good
3Average
2Below average
1Poor

The final score is the average of the graded tests (nine 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 can't export without a watermark or use a brand kit; I test them again on a paid plan when I can. A tool that cannot make one of the three briefs at all (a logo maker or a website builder, say) is graded on the briefs it can make, with the counts scaled to those briefs, and the review says which brief it could not take.

The 9 tests

1.Brief to First Draft

The clock starts when I submit the brief and stops when every piece in it exists as an editable design, not a preview. Each of the three briefs is timed from a fresh project; the median is the grade. Failed generations, retries and loading time count. The time to the first on-screen result is reported alongside.

  • 5: median under 5 minutes
  • 4: 5 to 9 minutes
  • 3: 10 to 19 minutes
  • 2: 20 to 44 minutes
  • 1: 45 minutes or longer, or a brief could not be completed

2.Brief Adherence

Each brief holds 20 required elements: the exact headline and body text, sizes in pixels, the five brand colours, the supplied images, the number of pages or screens, and layout rules such as the logo top left. Each is checked in the first draft before any editing. Invented content (text I did not supply, prices, reviews, stock photos used instead of the supplied ones) is counted separately.

  • 5: 58 to 60 right, nothing invented
  • 4: 54 to 57 right, at most one invented item
  • 3: 45 to 53 right, or two or three invented items
  • 2: 30 to 44 right, or invented reviews or claims
  • 1: under 30 right

3.First Draft Quality

Every piece of the three first drafts is opened at its required size and, for the app and the landing page, at 360, 768 and 1440 pixels wide. Faults are counted: text failing WCAG AA contrast (measured with a checker), text cut off or overflowing, overlapping elements, inconsistent spacing, no clear type hierarchy, placeholder text or images left in, images cropped through a face or a product, and tap targets under 44 pixels.

  • 5: 0 to 2 faults
  • 4: 3 to 5 faults
  • 3: 6 to 10 faults
  • 2: 11 to 20 faults
  • 1: 21 or more faults

4.Editing Freedom

Change a font across the whole design, change one brand colour everywhere, move an element by hand, resize a text box, swap an image and set its crop, edit one word without regenerating the block, add a new section or slide, align three elements with guides, change the canvas size and keep the layout, regenerate one element from a written instruction, undo three steps, and lock an element. Each counts when done without a workaround or support.

  • 5: all twelve
  • 4: 10 or 11
  • 3: 7 to 9
  • 2: 4 to 6
  • 1: 3 or fewer

5.Brand Kit & Consistency

I upload the shop's logo, two font files and five colour values as the tool's brand kit, or the nearest thing it offers. Then I generate ten new designs from short prompts (a sale post, an email header, a price list and so on) and check each one for the right fonts, colours from the kit only, the logo unaltered and no off-brand stock imagery. A design counts only when it needs no fixing.

  • 5: all ten on brand
  • 4: 8 or 9
  • 3: 6 or 7
  • 2: 3 to 5
  • 1: 2 or fewer, or no brand kit at all

6.Export & Handoff

Each point is tried on the real files: PNG at the exact brief size, SVG with editable text or shapes, print PDF with bleed, an editable file that opens in Figma or Sketch with layers intact, HTML and CSS or React code that runs on a plain host, fonts embedded or named in the export, full resolution with no upscaling artefacts, bulk export of a whole set, no watermark on the plan tested, and export of the source assets I uploaded.

  • 5: all ten
  • 4: 8 or 9
  • 3: 6 or 7
  • 2: 3 to 5
  • 1: 2 or fewer

7.Sharing & Collaboration

With a second account I try: a view-only share link, a comment pinned to an element, real-time editing by two people, version history with a restore, roles that stop an editor from deleting a project, a shared template or component, an approval or review step, and a team brand kit the second account can use but not change.

  • 5: all eight
  • 4: 6 or 7
  • 3: 4 or 5
  • 2: 2 or 3
  • 1: 1 or none

8.Rights & Commercial Use

I read the terms of service, the content policy and the plan page on the account I used, and I check the settings for a training opt-out. I record who owns generated designs, whether commercial use is allowed on the free and paid plans, whether uploaded files and brand assets are used to train models and how to opt out, and whether the vendor offers any indemnity for generated content.

  • 5: ownership, commercial use, opt-out and indemnity all clear
  • 4: ownership and commercial use clear, no indemnity
  • 3: commercial use only on paid plans, or opt-out hard to find
  • 2: terms unclear on ownership or training
  • 1: vendor claims rights over my designs or my uploads

9.Pricing & Value

Every plan is read from the checkout, not the marketing page. The figure graded is the yearly cost of the cheapest plan that allows commercial use, removes any watermark and includes a brand kit. I count how many generations the three briefs used and check whether that fits the plan's monthly credit or generation allowance. Overage prices, the renewal price against the first-year price and the payment methods seen at checkout are listed.

  • 5: under $120 a year, briefs fit the allowance, same price at renewal
  • 4: $120 to $199 a year
  • 3: $200 to $299, or renewal more than 50% higher
  • 2: $300 or more, or the briefs need paid top-ups
  • 1: poor value or hidden costs

Documented but not graded

Templates, stock libraries, AI image quality, animation, multilingual text and the mobile app 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 and real exported designs, never promotional descriptions or showcase galleries.
  • Identical briefs, tests, thresholds and sample sizes for every tool in the category.
  • Authorised accounts only; no password sharing, no bypassing of plan limits or bot checks, and test purchases only in test mode or refunded.
  • Every grade has evidence: a measurement, a screenshot, an export or a checklist, 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 design tool for testing. No AI design tool has been through these tests yet; every one I follow is listed on the AI design tools page, and reviews appear there as tests finish.