Test log
Banani test data
Raw results for the Banani review. The verdict is in the review.
- Tested
- 7 November 2026
- Tester
- Miles Foster
- Account
- Free demo account, signed up by RankDude, nothing paid to the brand
- Tests
- 9
- Screenshots
- 6
- Method
- Published methodology
- Environment
- Google Chrome on desktop
Results
| ID | Test | Key measurement | Score |
|---|---|---|---|
| T01 | Brand Kit & Consistency | – | n/a |
| T02 | Sharing & Collaboration | – | n/a |
| T03 | Brief to First Draft | Brief to draft: 49 seconds | 5/5 |
| T04 | Brief Adherence | Required elements right: 17 of 20 | 3/5 |
| T05 | First Draft Quality | Design faults: 7 | 3/5 |
| T06 | Editing Freedom | Edits that worked: 6 of 12 | 2/5 |
| T07 | Export & Handoff | Points that worked: 6 of 10 | 3/5 |
| T08 | Rights & Commercial Use | Owner of the output: You | 3/5 |
| T09 | Pricing & Value | Cheapest paid plan, monthly: $20 | 4/5 |
| Overall (in the review) | 3.9/5 | ||
Brand Kit & Consistency
n/aNot Tested- Method
- Upload logo, fonts and colours as a brand kit, then generate ten designs and check them.
- Remark
- Not graded; left out of the average.
Raw notes
Upload logo, fonts and colours as a brand kit, then generate ten designs and check them.
Not run: ten generations at about 3 credits each do not fit 12 credits, and the editor accepts no font upload.
- Needs a paid month.
Sharing & Collaboration
n/aNot Tested- Method
- Eight points tried with a second account.
- Remark
- Not graded; left out of the average.
Raw notes
Eight points tried with a second account.
Not run: no second account was made. A Share button and an Invite link exist.
- Needs a second test account.
Brief to First Draft
5/5Tested- Method
- The landing page brief for Pinecone & Thread was typed into the home prompt box with the logo and the yarn photo attached.
- Brief to draft
- 49 seconds
- Credits for the brief
- 3
- Remark
- Scale 5: median under 5 minutes. Only one of three briefs was run, so the median is that one run.
Raw notes
The landing page brief for Pinecone & Thread was typed into the home prompt box with the logo and the yarn photo attached. The clock ran from pressing Generate to the tool going idle.
The finished page appeared 49 seconds after I sent the brief and it was an editable design. The run used 3 credits, from 12 to 9.
For about 35 seconds the canvas showed a grey panel marked "Designing" while the status line named files such as LandingPage.jsx. The page was on screen by 40 seconds, and the tool kept tidying up until 49.
- One brief only: the app and bakery briefs did not fit the 12 credits.
Brief Adherence
3/5Tested- Method
- Twenty required elements of the landing brief were read from the first draft before any edit: sizes from the export, text, fonts and colours from the page's own code.
- Required elements right
- 17 of 20
- Invented items
- 6 kinds
- Remark
- Scale 3: 45 to 53 of 60 right, or two or three invented items. With one brief the thresholds are scaled to 20 elements, so 17 of 20 sits in that band.
Raw notes
Twenty required elements of the landing brief were read from the first draft before any edit: sizes from the export, text, fonts and colours from the page's own code.
17 of 20 were right: a page 1440 pixels wide, sections in the right order, the three header links, the hero headline, line and button, the three cards with names and prices on white, the classes title and line, the footer address and hours, the oat, bark, moss and berry colours, and Inter for body text.
Three failed: the headlines are set in PT Serif, not Fraunces (the tool offers no Fraunces); the header shows my logo rebuilt in code (a dark circle with a P and the shop name in PT Serif, the same shapes as the file I supplied), which was counted as missed because the file itself was not used; and my logo was enlarged and cut off at the edge of the hero, while the yarn photo was not used.
Content I did not supply was added: an email address, Add to bag buttons, a Book a spot button, a View all link, an Our yarns heading, product labels and three plain coloured placeholder blocks (tan, blue and khaki).
- One brief only. The first-draft picture is a 1x capture; later captures are 2x.
First Draft Quality
3/5Tested- Method
- The page was checked at 1440 pixels for contrast, cut-off text, overlaps, spacing, hierarchy, placeholders and crops.
- Design faults
- 7
- Remark
- Scale 3: 6 to 10 faults.
Raw notes
The page was checked at 1440 pixels for contrast, cut-off text, overlaps, spacing, hierarchy, placeholders and crops. Contrast was worked out from the hex values.
Seven faults. The hero line is 4.48 to 1 on the oat background where WCAG AA asks 4.5, and the footer email is 2.50 to 1 on dark brown. The hero shows my logo cut off, the classes photo repeats the same logo, and the three product cards hold plain coloured placeholder blocks.
Everything else held: no text cut off, no overlap, a clear headline and body order.
- Only the 1440 pixel layout was checked. 360 and 768 pixels need a mobile version, which is a new generation that I did not run.
Editing Freedom
2/5Tested- Method
- Twelve fixed edits on the generated landing page.
- Edits that worked
- 6 of 12
- Credits for the picture instruction
- about 7
- Remark
- Scale 2: 4 to 6.
Raw notes
Twelve fixed edits on the generated landing page. An edit counts when it did what was asked with a normal control and no workaround.
Six worked: the heading font changed everywhere from the Assets panel, one colour changed everywhere (berry to a dark red on every price and link), one word was retyped in place, a new screen was added, a written instruction replaced the hero picture, and three undo steps reversed my edits.
Six did not: dragging an element moved the whole page on the canvas, I found no control to resize a text box, align with guides, lock an element or set an image crop, swapping a photo needed a written instruction, and the canvas size menu offers only Desktop or Generate Mobile, a new AI generation.
The font list holds ten fonts and Fraunces is not one. The written instruction took 55 seconds and about 7 credits, from 9 to 2. The direct edits cost nothing.
- The new blank screen landed on top of the first screen on the canvas and was not filled, because filling it costs credits.
- I retook the image swap on 7 November: the toolbar, the right-click menu and a double-click on the hero picture were checked again and hold no replace or upload control, so that edit stays failed.
Export & Handoff
3/5Tested- Method
- Ten export points tried on the finished page: PNG at the brief size, SVG, print PDF with bleed, an editable file, code that runs on a plain host, fonts named, full resolution, bulk export, no watermark and export of the source assets I uploaded.
- Points that worked
- 6 of 10
- Remark
- Scale 3: 6 or 7.
Raw notes
Ten export points tried on the finished page: PNG at the brief size, SVG, print PDF with bleed, an editable file, code that runs on a plain host, fonts named, full resolution, bulk export, no watermark and export of the source assets I uploaded.
Six of ten worked: the PNG is 4320 pixels wide for the 1440 pixel page, a JPG of the same size, a zip of two screens, no watermark, Copy HTML/CSS gave 30 KB of code with a Google Fonts link naming the fonts, and the PNG shows no upscaling.
Four did not: there is no SVG and no print PDF, the Figma tab says "Figma export is a paid feature", and the photo I uploaded is only linked from Banani's storage, with no file to download. In the exported PNG the arrow in View all shows as a box.
- Figma export needs a paid plan and was not tried.
Rights & Commercial Use
3/5Tested- Method
- I read the terms of service, the privacy policy and the security page, reached from the links in the site footer, and looked in the settings for a training opt-out.
- Owner of the output
- You
- Training on free content
- Allowed
- Remark
- Scale 3: commercial use not stated, and the only way out of training is a paid plan.
Raw notes
I read the terms of service, the privacy policy and the security page, reached from the links in the site footer, and looked in the settings for a training opt-out.
Ownership is clear: "You retain ownership of the designs, prompts, and other content you create". The terms say nothing about commercial use of the output.
Training depends on the plan: "For accounts created on or after 23 June 2026, we may use de-identified User Content to train our AI models" and "We never use paid-plan content for training". I found no opt-out for a free account. The indemnity in the terms runs from the user to Banani, so Banani gives none for generated content.
- No lawyer's reading; only the pages above.
Pricing & Value
4/5Tested- Method
- Plans were read on the public pricing page in both billing views and on the credits page in the product.
- Cheapest paid plan, monthly
- $20
- Same plan, paid yearly
- $12 a month
- Remark
- Scale 4: $120 to $199 a year, on yearly billing.
Raw notes
Plans were read on the public pricing page in both billing views and on the credits page in the product.
The cheapest paid plan is Plus: $20 a month, or $12 a month paid yearly, about $144 a year. It carries 100 credits a month and 10 a day, and lists Figma and code export.
Free has 12 credits a month and 3 a day. The landing brief used 3 and one picture edit about 7.
- No plan was bought; renewal prices were not seen.
Other features checked
Coverage
Tested
- The landing page brief typed into the home prompt box with the logo and yarn photo attached
- Twenty required elements read from the page's code, with a contrast check
- Twelve fixed edits with the editor's own controls and one written instruction
- Ten export checks: PNG, JPG, a zip of two screens, Copy HTML/CSS and the Figma tab
- Terms of service, privacy policy, security page and pricing page read
Not tested or limited
- Only one of the three fixed briefs was run: the landing brief cost 3 credits and one image edit cost about 7, so a second brief would pass 85 per cent of the 12 credits. The grades for speed, adherence and quality rest on that one brief, scaled from 60 elements to 20
- The page was checked at 1440 pixels only: the 360 and 768 pixel checks need a mobile version, which is a new paid generation
- Brand kit was not graded: it needs ten new generations, which do not fit the credits
- Collaboration was not graded: every point needs a second account
- The first draft was replaced by the written image edit, so the first-draft picture is a 1x capture taken while the page was open
- No plan was bought, so Figma export and renewal prices were not tested
Still to test
- The app and bakery briefs on another day, when the credits refill
- The 360 and 768 pixel views through a mobile generation
- Brand kit with ten generations and Figma export on a paid month
- Collaboration with a second account