Short answer: The most reliable way to raise an SEO score is a cycle: fix a batch of related issues, wait until the changes are live, run a re-check and compare the result with the previous one. Starter, Pro and Agency include re-checks, and Pro adds comparison of audits over time. Work in batches of three to five fixes by shared cause, note each change, and stop a cycle only when the Top fixes list is short or the remaining items are not worth the effort.
A first report can feel like a mountain. Dozens of issues, seven area scores and a list of pages. The temptation is to fix everything at once, or to fix a little here and there without a plan. Both approaches waste effort, because you cannot tell which change worked, and you cannot tell whether a new problem appeared along the way. A repair cycle replaces guessing with evidence. This article describes the cycle step by step and shows how to use the features of Site SEO AI Audit in each step.
Why a cycle beats a one-off fix
A single audit is a snapshot. It tells you what is wrong today. It does not tell you whether your repair worked, whether it introduced a side effect or whether another change undid it. A cycle gives you a feedback loop. You make a change, you measure it, and the measurement guides the next change. This is how engineers work, and it works for SEO too.
The cycle also protects you from a common trap: fixing what is easy rather than what matters. The report ranks fixes by the points each would add to the score. In the sample report on the product’s home page, a canonical problem is worth nine points, duplicate titles five and blocked AI assistants two. A cycle that follows the ranking spends its first batch on the largest gains.
Step 1: choose a batch
Take the top of the “Top fixes by impact” list and group items that share a cause or a place. Three to five fixes per batch is a good size. Examples of natural batches:
- A template batch: titles, meta descriptions and headings that all come from one template or SEO plugin setting.
- A link batch: broken internal links, links to redirects and redirect chains, which are all cleaned in one pass through the menus and content.
- A crawl batch: robots.txt, sitemap and noindex settings, which are all controlled by a few files and settings.
- A page batch: thin pages, images without alt text and pages with weak internal links in one section of the site.
Too small a batch uses your checks quickly. Too large a batch hides the effect of each change. If you do not yet know where to start, the guide to fixing issues by impact explains how to read the list, and the article on prioritising SEO fixes adds general rules for choosing.
Step 2: make the changes and note them
Work through the batch. For each fix, the issue card gives the explanation, the repair and the affected pages, and WordPress sites get steps for wp-admin and the SEO plugin. While you work, keep a simple log with the date, what you changed and where. A line such as “12 Nov: title template for products now includes brand and category” is enough.
The log matters later. When the next check shows an unexpected result, you can look back and see what you touched. It is also the proof that you can show to a client, a manager or yourself. The article on SEO changelogs shows a layout that works.
Step 3: let the changes go live
Many failed re-checks are not failures of the fix but of timing. Page caches, CDN caches and browser caches can serve old versions of pages for minutes or hours. Before you re-check, clear every cache you control. Then open three pages that were affected and look at them: view the source for a title or a canonical, follow a link that you fixed, load the robots.txt. If the change is visible to you, it is visible to the crawler.
Step 4: re-check
Start a re-check from the report page. The product crawls your site again, with the same method, up to 10,000 pages. Starter, Pro and Agency include re-checks at any time, and each re-check uses one of the checks of your monthly allowance: 5 on Starter, 20 on Pro and 100 on Agency. A small site takes a few minutes, and large sites take up to about twenty minutes.
Plan the number of cycles to match your allowance. On Starter, a typical month might be one baseline, three repair cycles and one spare. The guide to choosing a plan by checks shows how to count.
Step 5: compare
Now read what changed. The total score is the headline, but the detail is more instructive. Look at the seven area scores, and at the list of issues: which have disappeared, which are smaller, which are new.
| What you see | What it usually means |
|---|---|
| The issue is gone, and the score rose | The fix worked; log it |
| The issue is smaller, not gone | The fix reached some pages; find the rest |
| The issue is unchanged | A cache, a wrong template or a different page type |
| A new issue appeared | The change had a side effect, or the crawler now sees more |
| The score fell | Check new critical issues first |
On Pro, comparison of audits over time does this reading for you: it shows the score change, the area changes, the new issues, the fixed issues and the issues that got worse. On Starter, do the same by hand with your saved reports, side by side. A detail worth knowing: the points in the Top fixes list are what the score would gain if that issue alone were fixed. When you fix several issues, the gain is close to the sum but not exactly equal, because the score is a weighted combination of area scores. Do not worry about a difference of a point or two.
When the score does not rise, or falls
Sometimes a good fix is followed by a lower score. The most common reason is that you opened the door. If a robots.txt rule or a noindex setting was hiding part of the site, removing it lets the crawler read more pages, and the new pages bring their own issues. That is an honest result: the problems were there all along, and now they are counted. Another reason is a side effect, such as a redirect rule that created chains. Look at the new issues in the comparison and treat them as the next batch.
A worked example of two cycles
Consider a regional furniture shop with about 600 pages and a first score of 63. The Top fixes list shows duplicate titles on the product pages, a missing sitemap and links to redirects in the menu. In cycle one, the owner fixes the product title template and adds a sitemap, then waits for the cache to clear and runs a re-check. The score reaches 70, the duplicate-title warning is gone and the sitemap issue has disappeared. A new notice appears: a few pages in the sitemap are not linked anywhere. In cycle two, the owner links those pages from their categories and updates the menu links that point to redirects. The second re-check shows 76. The log has two entries, each with a date, a change and a score. That record is worth more than any memory of what was done.
When to stop
A cycle has diminishing returns. The first batches usually give the largest gains, and later ones give a point or two. Stop when the Top fixes list is short, when the remaining items are notices that do not matter for your goals or when the effort is greater than the benefit. A score above 90 on a normal site is a strong result, and a perfect score is not a goal. Keep the rhythm going with a monthly check, and use Pro’s e-mail alerts when you want a warning if the score drops after a later check. For ideas about how often to repeat full checks, see the audit frequency guide.
To begin your own cycle, create a free account and run your check and choose a plan on the pricing page to unlock the report.
Related reading
- Fix SEO Issues by Impact: Using Your Audit Fix List
- SEO Changelog: Record Changes to Explain Traffic Moves
- SEO Testing: Measure Changes
The bottom line
Fix a batch of related issues, let the changes go live, re-check and read what changed. Log each cycle, treat new issues as the next batch and stop when further gains are small.
BUJ
Which plans include re-checks?
Starter, Pro and Agency. Each re-check uses one of your monthly checks.
How many fixes should one batch contain?
Three to five fixes that share a cause or a place. It is large enough to be efficient and small enough to see the effect of each change.
Why did my score fall after a fix?
Often the crawler can now read pages that were hidden before, and they bring their own issues. Check the new issues in the comparison and treat them as the next batch.
Do the points in the fix list add up exactly?
Not exactly. Each figure shows the gain from fixing that issue alone, so several fixes together come close to the sum.
Where is the comparison of two checks?
Pro and Agency include comparison of audits over time. On Starter, compare the saved reports side by side.


