Website analysis and questionnaire
Onboarding has two parts. First BoostUp reads your website and writes a report. Then you answer a short questionnaire about the things a website does not say. Both feed Brand Brain.
Start an analysis
There are two ways in:
- Without an account. Enter your website address on the homepage. The analysis runs right away and you get a report you can share.
- With an account. Choose New project in the project switcher and give it your website. The analysis starts when the project is created. You can run it again later from the project.
Only one analysis runs per project at a time.
What we read
The crawler reads your website the way a search engine does, with these limits:
| Limit | Value |
|---|---|
| Pages | up to 60 |
| HTML in total | up to 8 MB |
| Time per page | 20 seconds |
| Which pages | your site only: the same host and its www twin |
- robots.txt is respected. If your robots.txt asks crawlers like ours not to read the site, the analysis stops and says so. Pages your robots.txt blocks are skipped.
- Your sitemap is read. We read
sitemap.xml(or the sitemaps your robots.txt lists, and one level of sitemap index) to find pages that are not linked from the homepage. - Pages that need JavaScript are noted. If a page shows its text only after scripts run, we count it as a JavaScript-only page. Search engines index such pages late and most AI assistants never see their text.
- We skip files (images, PDFs, scripts, styles) and pages such as cart, checkout, login and account pages.
- A larger site is sampled: the report says when we stopped at the page or size limit.
Our crawler identifies itself with the user agent BoostUpBot/1.0 (+https://boostup.cloud/docs/onboarding). You can
address it by that name in your robots.txt.
We only connect to public internet addresses. Addresses on private or local networks are refused, only the standard web ports (80 and 443) are used, addresses with a username or password are refused, and every redirect is checked again before we follow it.
What we extract
Each fact we find is stored in Brand Brain with its source:
- Observed facts are read straight from your pages, and each one links to the page it came from: the business name, the description, the languages of the site, social profiles, contact details, products named in the page markup, and technology hints (for example WordPress, Shopify or a Google Analytics tag).
- Inferred facts come from an AI reading of the text we crawled: industry, products and services, target customers, markets, brand tone, key selling points, competitor candidates and a summary of your existing content. Each one carries a confidence and, when the model names one, the crawled page that supports it. Inferred facts never go above 85% confidence. A page link the model gives that is not one of the crawled pages is dropped.
The AI reading uses a low-cost extraction model and sees only text from your crawled pages. If AI extraction is not configured on the server, or the call fails, the analysis still succeeds with observed facts only, and the report says that the AI reading was skipped and why.
The free analysis without an account
- Start it from the homepage. You do not need an account.
- The report is kept for 24 hours. After that it is deleted.
- You get a share link to the report. Anyone with the link can open it for those 24 hours. Contact details found on your site are left out of the shared report.
- If you create an account in the same browser within the 24 hours, the analysis becomes your first project, with its facts already in Brand Brain. This works whether the analysis has finished or is still running.
- Free analyses are limited per day: 5 from the same visitor, and a daily total across all visitors. When a limit is reached, the page asks you to create a free account or to try again later.
The website report
The report lists what we found on the pages we read. In the app it is Website report in the sidebar.
Health checks
| Check | Rule |
|---|---|
| HTTPS | The site answered over https:// |
| robots.txt | A robots.txt file was found |
| Sitemap | A valid sitemap was found, with the number of addresses in it |
| Missing titles | Pages with no title tag |
| Duplicate titles | Pages whose title is the same as another page's (ignoring case) |
| Missing meta descriptions | Pages with no meta description |
| Missing H1 | Pages with no H1 heading |
| Thin pages | Pages with fewer than 250 words of text (JavaScript-only pages are counted separately) |
| JavaScript-only pages | Pages with fewer than 80 words of text and an empty app container that scripts fill in |
| Broken internal links | Links to pages on your site that answered with an error (HTTP 400 or higher), with the page that linked to them |
| Unreachable pages | Pages on your site we could not load at all |
Each check shows the count and up to 10 example pages. The report also shows how many pages we read and any limits we hit.
Suggested next steps
The next steps are fixed rules over the counts above. There are no scores and no estimates: each step names the problem we counted.
| Priority | Step | When |
|---|---|---|
| High | Serve the site over HTTPS | The site answered over plain HTTP |
| High | Make the main content readable without JavaScript | At least one JavaScript-only page |
| High | Fix broken internal links | At least one broken internal link |
| High | Add page titles | At least one page without a title |
| Medium | Publish a sitemap | No sitemap found |
| Medium | Give every page its own title | At least one duplicate title |
| Medium | Write meta descriptions | At least one page without a meta description |
| Medium | Add one H1 per page | At least one page without an H1 |
| Low | Expand thin pages | At least one thin page |
| Low | Add a robots.txt | No robots.txt found |
| Low | Keep the basics as they are | None of the checks found a problem |
The questionnaire
After the analysis, BoostUp asks 10 short questions. Your answers are saved as owner facts in Brand Brain. Owner facts always outrank what the crawl found, and a later analysis never overwrites them.
| Step | Question | Answer |
|---|---|---|
| 1 | What you sell and to whom | Free text |
| 2 | Your three best customer types | One to three entries |
| 3 | What you want more of | One or more of: leads, bookings, online sales, local visits, hiring, brand |
| 4 | Markets and languages | At least one market and one language |
| 5 | Monthly ad budget appetite | One of: none, under 500 EUR, 500 to 2000 EUR, over 2000 EUR per month |
| 6 | Brand voice | Three sliders: formal to casual, calm to bold, technical to simple |
| 7 | Claims we must never make | A list, can be empty |
| 8 | Competitors you admire | A list, can be empty |
| 9 | Accounts you have | Any of: WordPress, Search Console, GA4, Meta, Google Ads, other (with a note) |
| 10 | Anything else your AI employee should know | Free text, can be empty |
- Saved per step. Each step is saved when you move on, so nothing is lost if you leave.
- Resumable. When you come back, the questionnaire opens at the first step you have not answered or skipped.
- Skippable. You can skip a single step or the whole questionnaire for now. While it is not complete, the project overview shows a reminder.
- You can change an answer later in Brand Brain. Editors and above can answer the questionnaire.