Swiss Knife SEO
Customer login Scan my site free

Home / How it works

A closed loop from finding to verified repair

Find the issue, rank it with site evidence, publish the approved change, and verify the outcome on the following crawl. The work stays visible from start to finish.

Create my free workspace Scan my site free
The core loop

Four steps, and the last one completes the circle

Every finding goes through the same four phases. Nothing is left as a number in a table and nothing is marked as done until a new crawl agrees that it is done.

1

Find

The crawler renders your pages using a real browser, just like a search engine, and then rates the site from 0 to 100. The results are ranked by impact and template reach, so a single meta rule affecting 400 pages ranks above a one-time typo. You see the exact URLs behind each number, not a bucket average.

2

Plan

Ask AI to provide a repair plan built from your own crawl data: the affected URLs, the crawler reports, and the page titles behind the issue. This means the plan conforms to your CMS and templates, rather than returning general advice that you still need to translate.

3

Publish

Take a copy-paste patch for titles, metas, robots statements, canonicals or security headers and place it in your stack. Or post the change to the margin CDN, where the answer will be rewritten on the fly without touching your CMS, and reverted with one click if you change your mind.

4

Check

At the next crawl, the exact URLs is checked again against the rule that the fix had to meet and the issue is marked as confirmed, noting the site scan that confirmed it. Owners, due dates, and SLA timers travel with each task, so work is completed based on evidence, not a commitment.

Renders JavaScript Organized according to impact and reach Grounded in your URLs Copy and paste or edge publishing Restore with one click Automatic re-verification Ownership and SLAs
The differentiator

Why a body link and a footer link are not the same

A normal crawl export counts internal links and stops there. Our method reads where each link is located and calculates a position-weighted internal PageRank for each page, since a link in the middle of a paragraph carries far more weight than the same link repeated in the footer of each template.

Orphaned pages

Pages that no other page links to in the text are outside the flow of internal justice. If you're just counting links, they can be completely missed because a standard menu technically links to them. Position weighting exposes them for the starved sites that they are.

Boilerplate-only pages

Some pages can only be accessed through navigation and footer links that are repeated throughout the site. They appear to be linked together in a flat export, but bring in almost no real equity. We label them so you can add a real in-body link from a relevant page.

Low equity sites

Your category and product pages should be strong, but many are on the lower end of the internal stock map. The report highlights the high-quality pages that are underlinked and identifies the source pages and anchors that would resolve the issue.

The practical advantage: The same crawl that finds these gaps also creates a ranking of the internal links to be added, each with the source page and the anchor text to be used. It's the difference between knowing a site is weak and knowing the exact three links that would strengthen it.

Read the score honestly

A low first score is a baseline, not a judgment

The Health Score is calculated from weighted issues per crawled page and therefore cannot be improved by crawling fewer URLs. A real site with real history usually starts lower than people expect.

A first crawl usually lands somewhere between 20 and 40 out of 100. This is normal, and it is the starting point from which you improve, not a judgment on your work. Because results are grouped by template, deleting a rule often raises hundreds of pages at once, so the number changes in real increments. The key value is the trend across crawls, and each point on it points directly to the pages behind it.

  • Rated per page crawled, so honesty is maintained even as the site grows
  • Thematic sub-ratings for crawlability, linking, on-page and security show where the number comes from
  • Each point extends to the exact URLs, so there is nothing hidden in an average
  • Template-level corrections move the score in visible jumps between crawls
Frequently asked questions

Before executing the loop

Do I need to install something to run the loop?
No. Add your URL and the crawl will run in the cloud with a real browser, so you don't need to install or update a desktop app. Publishing a copy-paste patch inserts a snippet into your own CMS or server, and Edge Publishing rewrites the response on CDN without even touching your CMS. Search Console and GA4 only connect when you need traffic context, and both are read-only. Edge publishing is the other connection and uses your own CDN credentials, so the rewrite occurs within your infrastructure.
What does the first crawl typically achieve?
A first crawl usually lands at 20 to 40 out of 100. This is the basis for improvement, not a judgment on your website. The score is calculated based on the weighted issues per page crawled. So it cannot be improved by less crawling. The number that matters is the trend across crawls when you clarify results by submission.
How is a fix marked as verified?
Each fix maintains the exact URLs and rule that must be satisfied. During the next crawl, these URLs will be checked again against this rule. If they are successful, the issue will be marked as verified with the crawl number that confirmed it. Owners, due dates, and SLA timers change with each task, so nothing is tied to a promise.

Check out the loop on your own website

Run a crawl, rank the results by impact, and watch as the first repair is verified. The free plan includes the full report suite for up to 100 URLs.

Create my free workspace Scan my site free