What it is
A small Python tool that Thraex wrote itself, built for WooCommerce stores. It does not log in, click buttons or change anything in the store: it only makes ordinary page requests, like a visitor's browser. The work needs no admin access. Admin is needed only once the fixes begin.
How it works
- The crawler takes the store's sitemap and knows which pages are products, categories, brands or plain pages.
- It visits the pages at a slow pace and reads the SEO fields of each: title, description, headings, images, internal links, price and stock.
- Rules look for missing and duplicate texts, broken redirects, wrong canonical addresses and pages that no internal link points to.
- Findings are grouped into 13 fix packages. The order comes from impact, effort and how many pages the problem touches.
- After the fixes the tool crawls the changed pages again and shows what got fixed.
What it checks
- Titles and descriptionsMissing, too long or too short, duplicated, and cut-off automatic texts.
- Product pagesDescriptions that are too short, product names that are only a code, missing reviews and markup, English text on an Estonian page, and colour variants sold as separate products.
- Categories and brandsText that is too short, empty pages and a default sort from the cheapest.
- Internal linkingPages that no internal link points to, and pages six or more clicks away from the home page.
- Duplicate contentIdentical and nearly identical texts on several pages.
- Technical errorsRedirects, canonical, slow responses and images without alt text.
- Cart cacheSigns that a stranger's shopping cart may be sitting in the page cache.
How it stays gentle on the store
- Slow paceAt most 2 requests at a time, half a second between them, and automatic slowing when the site slows down.
- Says who it isThe store's log shows ThraexSEOAudit and a “read-only SEO audit” note. The crawler does not hide, and it reads and follows robots.txt.
- Risky addresses are skippedCart, checkout, customer account, admin and search are left out.
- Stops by itselfIf a fifth or more of the store's answers are blocks, the crawler stops and does not retry.
- The computer is protectedA separate guard pauses the work if the computer's load, memory, disk or battery requires it.
- Nothing personal is keptFull page text, headers and tokens are not stored.
An example from a real store
A full crawl of a large online store (AirsoftGO) went through without the store blocking it. Among other things it found that 80% of product pages had the default title template, 99% of categories opened sorted from the cheapest and half of the products had a description under 50 words. All of it is merged into one ranked list.
An honest status
- The tool is still being testedOne full crawl has been run on one real store. It has not been tried on another store, and for a different theme the page-reading rules need to be rechecked.
- Some parts are uncheckedFinding reviews and short descriptions is not proven. We make no claims about them until we have checked them by hand.
- No JavaScript contentThe crawler does not render pages, so anything that only appears in the browser goes unnoticed.
- The order is an aidThe list weighs only impact and page count, so a serious but rare error needs a separate look.
- No promised resultsThe crawler finds errors. Whether and how many extra Google clicks follow shows in Search Console numbers, not in a promise.
Get started
Write to margus@thraex.ee and include the store's address. We will tell you whether the crawler suits your store and how long a crawl would take. Before a full crawl we let you know, so the store's log holds no surprise.