SEO Score History Example: Trend Charts Across 12 Weeks
This example shows the aiwebpageseo Score History Tracker for a SaaS marketing site tracked weekly for 12 weeks. AEO score improved from 42 to 78 after adding FAQPage schema and llms.txt. PageSpeed improved from 61 to 84 after hero image preloading. E-E-A-T went from 54 to 69 after adding author bios.
| Week | Date | AEO | E-E-A-T | PageSpeed | Site Audit | Notes |
|---|---|---|---|---|---|---|
| Wk 1 | 2026-01-05 | 42 | 54 | 61 | 67 | Baseline |
| Wk 4 | 2026-01-26 | 49 | 58 | 68 | 71 | FAQPage schema added ↑ |
| Wk 7 | 2026-02-16 | 62 | 63 | 76 | 79 | Hero image preloaded ↑ |
| Wk 9 | 2026-03-02 | 70 | 67 | 81 | 85 | llms.txt + author bios ↑ |
| Wk 12 | 2026-03-21 | 78 | 69 | 84 | 88 | Latest ✓ |
The AEO score went from 42 to 78 after FAQPage schema and an llms.txt file were added. It is worth being precise about what happened there, because the obvious reading is circular.
The AEO score is computed from a checklist, and two items on that checklist are "has FAQPage schema" and "has llms.txt". Adding those files therefore raised the score with certainty, immediately, and would have done so even if no assistant on earth ever fetched either one. The chart is not showing you that the changes worked. It is showing you that the scorer noticed them.
That does not make the score useless — a checklist is a fine thing to have — but it means a score chart is a record of what you did, not a measurement of what it achieved. Evidence of achievement lives elsewhere: crawler hits in your access logs, impressions and clicks in Search Console, actual citations. Chart those beside this and you will see how loosely they move together.
"When did this start?" is the most valuable question in a regression
A metric moves, nobody remembers shipping anything, and the debate about the cause runs for a week. A trend line ends that debate: the inflection has a date, the date has a deploy, and the deploy has a diff. This alone justifies keeping the history, and it is why the chart is worth more in the bad weeks than the good ones.
Did the fix stick?
Technical fixes regress silently — a plugin update, a theme change, a CDN rule. A score that rose in week 3 and slid back by week 9 is a fix that came undone, and nothing else in your stack would have told you. Watching for the slide is the point of the series; celebrating the rise is not.
Note what neither of these requires: an interpretation of why a score is what it is. The trend is a timeline of events on your own site. It becomes misleading precisely when it is read as a measurement of how the world responded.
| Score | What it really is | Relationship to outcomes |
|---|---|---|
| PageSpeed 61 → 84 | A lab simulation on a synthetic device | The most defensible of the three. Core Web Vitals are a genuine ranking consideration, though a small one — a tie-breaker between comparable pages. Google uses field data, not this lab score, and the two diverge routinely. The stronger argument is that people abandon slow pages. |
| AEO 42 → 78 | Checklist compliance | Measures whether you have the files and markup the checklist looks for. No assistant operator publishes retrieval logic, so no tool can connect this to citations. Treat it as a readiness list. |
| E-E-A-T 54 → 69 | A model of rater-guideline characteristics | There is no E-E-A-T score. It is criteria used by Google's human quality raters, not a value computed for your URLs, and no schema property confers it. Author bios are worth adding because attribution should be real, not because a dial moved. |
All three rose. All three rose because you did the things the scorers look for. Whether the site is better is a separate question, and the answer to it is mostly in the content, which none of these three measures at all.
Scores fall for three reasons and only one of them is a defect. Confusing them produces a particularly damaging kind of busywork: undoing good work to make a number recover.
The real case
A deploy removed the schema, a plugin stopped emitting a field, a WAF rule started refusing a crawler. Investigate, fix, and add a check so it cannot happen silently again.
The measurement moved, not the site
Scoring methodologies get updated, thresholds shift, a new check is added. Your site is unchanged; the ruler is different. Nothing to fix, and a chart that spans the change will show a step that looks alarming and means nothing.
Ratios fall when you add substance
Several composite scores include coverage and density ratios. Publish thirty genuinely useful new pages without markup on all of them, and your schema-coverage percentage drops even though the site is materially better. Keyword density behaves the same way — it is a ratio that rises when you delete content, which is why it was never worth optimising in the first place.
The rule: always find out what happened before you act on a number. A score is a summary, and summaries are exactly where causation goes to hide.
Every line on this report is generated from your own site. That is the structural limitation, and it is why a second chart belongs next to it: one showing how the world responded.
Pull impressions and clicks from Search Console over the same twelve weeks and overlay them. What you will usually find is that the two sets of lines are only loosely related, and the divergences are the interesting part. A period where the scores climbed and impressions did not is a period where you satisfied a checklist and changed nothing that mattered. A period where impressions climbed with no movement in the scores is a period where somebody wrote something good, and no technical metric noticed.
Add crawler evidence to the same view if you can. Whether OAI-SearchBot, ClaudeBot and PerplexityBot are actually reaching your origin and receiving 200s is a fact recorded in your access logs — and it is the one precondition that no score on this page can verify. A perfect AEO score sitting behind a WAF rule that silently drops requests from unfamiliar user-agents is a score describing a site that assistants cannot read at all.
The discipline is simple to state and hard to keep: measure the site with scores, measure the outcome with logs and Search Console, and never let the first stand in for the second. A team that optimises the score will get a better score. A team that watches both will find out, eventually, which of their work was real.
Any number that gets tracked weekly and shown to a manager will eventually be optimised for its own sake. Three habits keep that from happening here.
Write down what you shipped, on the date you shipped it
A trend line without annotations is a shape. A trend line with "llms.txt added" and "hero preloaded" against the right weeks is an argument you can check. It also stops the most common failure, which is a team confidently attributing a movement to whichever change they most wanted to have worked.
Two changes in one week means neither is measurable
The 42-to-78 jump followed two changes at once, so the chart cannot separate them. If you care which one did the work — and here, since both are checklist items, neither did anything outside the checklist — ship them apart and leave a gap.
Use scores to catch regressions, not to chase 100
The last twenty points of any composite score are the least valuable and the most expensive, and they usually reward adding markup that describes nothing. Decide what "not broken" looks like, alert when it slips, and spend the recovered time on the pages themselves — which is the only lever on this page with unbounded upside.
How do I save scores to history?
After running any audit — AEO, E-E-A-T, PageSpeed, Readability or Site Audit — click the Save to History button in the results. Scores are saved locally in your browser and synced to your account on paid plans. You can save multiple URLs and compare them on the Score History page.
How far back does score history go?
Free plan stores the last 10 score saves per tool. Bronze and Silver plans store 90 days. Gold and above store unlimited history. CSV export is available on all paid plans.
Can I track multiple URLs?
Yes. Each saved score is associated with the URL that was audited. On the Score History page, use the URL filter to switch between tracked pages. Competitor comparison allows you to overlay your URL against a competitor's score trend on Gold plan and above.
Did the AEO score rise because of the FAQPage schema and llms.txt?
The chart shows that the score rose after those changes; it cannot show that they caused the rise, and the distinction is not pedantry. The AEO score is computed from a checklist, and adding FAQPage schema and an llms.txt file adds ticks to that checklist — so the score was guaranteed to go up the moment those files existed, whether or not any assistant ever read them. That is a tautology, not a result. If you want to know whether the changes did anything in the world, the evidence has to come from somewhere else: crawler hits in your access logs, impressions in Search Console, actual citations. A score improving because you satisfied the thing that computes the score is not evidence of anything.
What can a 12-week trend line legitimately tell me?
Two things, both useful. It tells you when something changed, which is invaluable when a metric moves and nobody remembers shipping anything — the date on the chart is where you start looking. And it tells you whether a change stuck: a score that rises and then decays is a fix that regressed, which is the single most common outcome of technical SEO work and the one nobody notices without a trend. What it cannot tell you is why anything moved, because these are scores computed from your own site rather than measurements of how the world responded to it.
Should I be worried when a score goes down?
Look at what changed before you look at the number. Scores fall for three quite different reasons: something genuinely broke, the tool changed how it measures, or you added content and thereby diluted a ratio. The third catches people out constantly — several composite scores include density and coverage ratios that fall when you publish more pages, so a site can improve materially and score worse. Investigate every drop, but investigate it by finding out what happened, not by taking action to make the number recover.
Which of these trend lines actually correlates with traffic?
None of them reliably, and any tool implying otherwise is overclaiming. PageSpeed has the most defensible link, because Core Web Vitals are a genuine ranking consideration — though a small one, acting as a tie-breaker between comparable pages rather than lifting a weaker page above a stronger one, and measured on field data rather than the lab score in a tool. The AEO and E-E-A-T scores are checklist compliance measures: E-E-A-T in particular is criteria used by Google's human quality raters and is not a value computed for your pages by anyone. If you want a line that tracks outcomes, chart impressions and clicks from Search Console alongside these, and watch how loosely they move together.
Related Demo Reports
Run Score History Tracker on Your Own Site
Get your real audit report with specific issues, fixes and actionable improvements — free to start.