Score History Fixes: What Your SEO Score Trend Is Telling You
A single audit score is a snapshot of one day. Score History and Trends collects the scores from every audit you run, charts them over time and flags the patterns worth your attention: significant drops, long plateaus, unusual swings and anomalies, along with links between your site deployments and changes in the score.
The shape of the line usually tells you more about what went wrong than the number itself. This page explains how to read each pattern, what normally causes it, and which guide walks you through the fix.
Reading the shape of the line
Different problems leave different fingerprints on a score chart. Before opening any audit detail, look at the shape over the last few months and match it to one of these patterns.
| Shape | What it looks like | Usual kind of cause |
|---|---|---|
| Cliff | A sharp fall between two audits | One change: a release, a plugin update, a robots.txt or template edit |
| Slope | A small fall almost every week for months | Slow decay: links breaking, pages ageing, images growing, scripts piling up |
| Shelf | Flat for months despite work going in | Easy fixes done; what remains needs bigger structural work |
| Sawtooth | Up and down every audit within a band | Different pages sampled, unstable pages, or frequent deploys |
| Bounce-back | Rises after a fix, then falls again | The fix was overwritten, or the problem had a second source |
| Annual dip | A fall at the same time each year | Seasonal content, campaigns or site changes repeating yearly |
Matching the shape first saves time. A cliff sends you looking for a single event on a specific date, while a slope tells you there is no single culprit and a steady maintenance routine is the answer.
The fix guides, by pattern
Falls
Fixing sudden score drops is the step-by-step diagnosis for a cliff: find the date, list what changed, compare audits before and after, and confirm the cause before fixing it. Fixing gradual erosion covers the slope, where stale content, accumulating broken links and creeping page weight each take a little off every week, and sets out a regular audit-and-refresh routine to stop it.
Stalls and instability
Fixing score plateaus explains why progress stops once quick wins are gone, and what bigger pieces of work tend to move the score again. Fixing high volatility helps you work out whether swings come from which pages were sampled, from frequent deploys or from genuinely unstable pages. Fixing regression after a fix covers fixes that do not stick, and how to check that a fix survives the next release.
Context and interpretation
Fixing benchmark drift is for when your own score is steady but competitors are improving faster, so you are falling behind without anything getting worse. Fixing seasonal pattern misreading explains comparing with the same period last year rather than last week. Fixing score and traffic divergence covers what it means when the score and your traffic move in opposite directions.
By type of site
What to track depends on how the site earns. Content sites link the score to publishing pace and track sections separately. Ecommerce tracks by category and around launches and peak seasons. B2B SaaS separates marketing pages from product and sign-up pages and watches the pages that lead to demos and trials.
Linking score changes to releases
Most sudden changes to a site's technical health are self-inflicted. That is good news, because a problem you caused is one you can find and undo. The tool surfaces correlations between deployment events and score shifts to help with exactly this.
Changes that most often knock the score down after a release include:
- A staging
noindextag or password protection accidentally carried over to the live site. - robots.txt rules edited or replaced by a platform update.
- A theme or plugin update that removes structured data, changes heading structure or alters canonical tags.
- New scripts, such as chat widgets, tag manager additions or tracking, that slow every page.
- URL changes without redirects, creating broken internal links across the site.
A correlation is a lead, not proof. If a release and a score drop happen on the same day, compare the audit findings before and after: the specific checks that changed should match what the release touched. If they do not, keep looking.
What a health score can and cannot tell you
A site health score summarises how well your pages meet technical and on-page checks at the time of each audit. It is a useful measure of things you control. It is not a measure of rankings or traffic, and treating it as one causes most of the confusion people have with score trends.
- It reflects the pages audited. If different audits cover different pages, part of any movement comes from the sample, not the site. Keep audit settings consistent when you want to compare over time.
- It does not see demand. Fewer people searching for your topic lowers traffic without touching the score.
- It does not see competitors directly. If others improve faster, your rankings can slip while your score holds steady. That is benchmark drift.
- It does not see search engine changes. When Google updates how it ranks pages, traffic can move with no change on your site. The Algorithm Impact fixes page covers that situation.
Use the score to keep your own house in order, and use traffic, rankings and conversions to judge whether the market is responding. The two sets of measures answer different questions, and the most useful reviews put them side by side: a healthy score with falling traffic points outside the site, while a falling score with steady traffic is an early warning worth acting on before customers notice anything.
Which patterns to deal with first
- Cliffs. A sudden drop often means something is actively broken, such as pages blocked from crawling. Investigate the same day.
- Bounce-backs. A fix that keeps being undone wastes effort every release. Find out how it is being reverted and put a check in place.
- Divergence where traffic is falling. If the score is fine but traffic is dropping, the cause lies outside what the score measures and needs a different investigation.
- Slopes. Not urgent on any given day, but the damage accumulates, so set up regular maintenance.
- Plateaus and benchmark drift. Strategic questions about where to invest next.
- Sawtooth and seasonal patterns. Often need understanding rather than fixing. Confirm the cause before changing anything.
When two patterns overlap, such as a slope with a cliff partway down, deal with the cliff first. It has a single cause you can find and reverse, and once it is gone the underlying slope becomes much easier to see and measure.
Order of work for tracking scores well
- Run audits on a regular schedule with the same settings, so the trend is comparing like with like.
- Open Score History and let a baseline build up over several weeks before drawing conclusions.
- Keep a simple log of releases, plugin updates, content changes and campaigns.
- When the tool flags a change, match the shape of the line to a pattern above, then open the matching guide.
- Compare individual findings from the audits either side of the change to find exactly what moved.
- After fixing, watch the next few audits to confirm the fix has held.
- Once a quarter, look at the year-long view alongside traffic and competitor data.
For what individual checks mean, see the Site Audit Guide and the Site Audit fixes. Speed-related movements are explained in the Core Web Vitals Guide, and the Competitor Analysis Guide and Backlinks Guide help when the answer lies with competitors or authority.
1. My score dropped overnight. What should I check first?
Check whether anything changed on the site in the day or two before the drop: a release, a plugin or theme update, a robots.txt edit or a change of hosting. Then compare the audit findings from before and after the drop to see which checks failed. The most damaging causes, such as a noindex tag or a robots.txt block copied from a staging site, are usually visible straight away and quick to undo once found.
2. Why does my score go up and down every week when I have not changed anything?
Several things can cause that. Audits may cover slightly different pages each time, some pages may load inconsistently because of slow servers or third-party scripts, and outside services your pages rely on can be briefly unavailable. Keep audit settings the same, check whether the same pages fail each time, and look at speed measurements on the affected pages. The volatility guide explains how to separate sampling noise from genuine instability on the site.
3. Is a higher score always better for rankings?
A better score means your pages pass more of the technical and on-page checks, which removes obstacles to good rankings. It does not guarantee them. Rankings also depend on how relevant and useful your content is, how many trusted sites link to you and how strong competing pages are. Think of the score as a measure of whether your site is in good working order, not as a prediction of where you will appear.
4. My score has been stuck for months. What am I doing wrong?
Probably nothing wrong. Plateaus are normal once the quick, template-level fixes are done. What remains is often harder: speeding up a heavy theme, restructuring navigation, improving thin sections of content or cleaning up old redirects. Look at which findings are still failing across many pages and group them into a small number of larger projects. Moving the score from a plateau usually takes one of those projects, not more small tweaks spread thinly.
5. My score went up but my traffic went down. How is that possible?
The score measures the condition of your pages; traffic depends on searchers, competitors and search engines too. Traffic can fall because fewer people are searching, a competitor has published something better, AI answers are satisfying the query before a click, or a search engine update has changed how pages are ranked. Check each of those before assuming the score is wrong. The divergence guide takes you through them in a sensible order.
6. How often should I run audits for the trend to be useful?
Regularity matters more than frequency. Weekly audits suit most small and medium sites and give a clear trend within a couple of months. Busy sites with frequent releases benefit from more frequent checks, particularly straight after deployments. Very small sites that rarely change can manage with less. Whatever you choose, keep the same settings each time, because a change in what is audited will look like a change in the site itself.
7. Should I compare this month with last month or with last year?
Use both, for different questions. Month on month tells you whether recent work or recent problems are having an effect. Year on year removes seasonal patterns, such as a shop adding and removing Christmas ranges or a school-related site going quiet in summer. If a dip appears at the same time every year, compare with the same weeks last year before deciding something has gone wrong. A dip that matches last year's is usually expected rather than a fault.
Score History and Trends collects scores from every audit you run, charts them over time, flags significant drops and plateaus, identifies anomalies and highlights links between site deployments and score changes.
Run the Score History →