---
title: "90-day SEO audit: what changed and what didn't | Aristral"
description: "Our 90-day SEO audit of aristral.com: what we changed, what moved in Search Console and on Maps, what got worse, and what we are doing next."
url: "https://aristral.com/blog/90-day-audit-what-moved-our-rankings-and-what-didnt-transparency-post"
canonical: "https://aristral.com/blog/90-day-audit-what-moved-our-rankings-and-what-didnt-transparency-post"
---

1. [Home](https://aristral.com/)
2. [Blog](https://aristral.com/blog)
3. [SEO](https://aristral.com/blog/category/seo)

# We audited our own SEO for 90 days: what moved and what didn't

Taha Bilal reports what changed in Aristral’s 90-day SEO audit, what moved in search, what did not, and what the team is doing next.

Taha Bilal 26 Apr 2026 6 min read

On this page
* [What we changed](#what-we-changed)
* [What moved in search](#what-moved-in-search)
* [What did not move, and what got worse](#what-did-not-move-and-what-got-worse)
* [What I think the evidence says](#what-i-think-the-evidence-says)
* [What we are doing next](#what-we-are-doing-next)

## Key takeaways

The review shows that technical cleanup improved crawl hygiene, Bristol profile actions rose, while recent Search Console clicks and average position fell. National rankings and AI citations remain absent in the checked sample. We are checking indexation and mobile performance, building permissioned evidence, and measuring again without assigning causes the data cannot prove.

**Written by Taha Bilal, co-founder of Aristral. Updated 20 October 2026.**

We audited our own site on 6 July, 14 September and 20 October 2026\. The work included crawl and index checks, Search Console, local profile data, live search results and a repeated site-quality score. It is a record of changes and outcomes; it cannot show that a particular technical fix caused a ranking change.

## What we changed

The July audit scored 47 out of 100, using an older rubric. That score is not directly comparable with the later scores. In September we switched to a seven-dimension rubric. The site scored 65 in the morning of 14 September. We deployed a set of fixes, then the team that made them re-scored the site at 82 that evening. On 20 October an independent re-measure, applying stricter checks, scored it at 76\. The independent score is the more useful current assessment. The same-day 82 was a self-assessment, so the six-point difference is not evidence that the site lost six points of quality in five weeks.

A large part of the September work was crawl hygiene. We found 54 programmatic city pages sitting at an average UK position of 40 or worse while taking 56% of UK impressions and producing about one click. We set about 590 city pages to noindex,follow and removed them from the main sitemap. A temporary cleanup sitemap gave Google a way to revisit those URLs. Impressions on that noindexed set fell from 4,756 in the 28 days to 20 September to 847 in the 28 days to 18 October. That is evidence that Google is dropping many of those pages from results. It is not evidence that our other pages gained those impressions or clicks.

The September production changes also removed several sources of crawl waste. Links from indexable pages into noindexed pages went from 938 to zero. Internally linked 404s went from 20 to zero at that check. Links that passed through redirects went from 839 to zero. FAQPage markup was reduced from 176 pages to one, and priceRange in structured data from 247 pages to zero. We removed dates that predated the company's incorporation from 34 blog posts. Median HTML size fell from 313 KB to 119 KB, and the main sitemap went from 247 to 222 URLs.

These are real maintenance wins: the site is cleaner to crawl and the markup more restrained. On their own they do not make the pages more authoritative or useful than competitors'.

## What moved in search

Search Console recorded 231 clicks and 180,042 impressions in the 90 days to 18 October. The most recent 28-day period, 21 September to 18 October, had 54 clicks, 55,988 impressions and an average position of 24.7\. The previous 28 days, 24 August to 20 September, had 89 clicks, 61,123 impressions and an average position of 18.3\. Clicks and impressions both fell, and average position worsened.

Profile actions on our Bristol Google Business Profile, previous 90 days and the 90 days to 17 October 2026\. Source: Google Business Profile Performance data for aristral.com.

| Item                                  | Value |
| ------------------------------------- | ----- |
| Website clicks, previous 90 days      | 8     |
| Website clicks, 90 days to 17 Oct     | 38    |
| Direction requests, previous 90 days  | 61    |
| Direction requests, 90 days to 17 Oct | 122   |

The timing matters. Our daily average position was strongest from 8 to 13 September, between 12.9 and 15.8, then worsened from 14 September. Google's September 2026 spam update began on 24 September. Those dates overlap with the decline, but overlap does not establish cause. Search Console query rows explain only 52 of the 231 clicks; the rest are anonymised. With a small click base and incomplete query visibility, a neat causal story would be misleading.

There was one clear local improvement. On our Bristol Google Business Profile, the 90 days to 17 October compared with the previous 90 days show website clicks rising from 8 to 38, direction requests from 61 to 122, and calls from zero to two. The profile had 14 reviews, all rated five stars, and all had received replies. This is a useful direction of travel for the place where we have a real presence, but these are profile actions, not confirmed leads or revenue.

The live search check on 20 October was mixed. We ranked fifth for “AI agency Bristol” and eighth for “AI automation agency Bristol”, both also appearing in the local pack. We were eighth for “n8n pricing” two days after publishing that guide. We did not appear in the top 30 for any of the 16 sampled national AI and automation agency terms. AI Overviews appeared on 20 of the 27 checked results, and none cited Aristral.

## What did not move, and what got worse

The cleanup did not produce national visibility for our commercial AI and automation terms. Nor did it make us a source cited by AI Overviews in the sample. A clean crawl can remove obstacles, but it cannot supply independent references or show a buyer what happened in a real project.

Search Console clicks for aristral.com: 24 August to 20 September 2026, and 21 September to 18 October 2026\. Source: Google Search Console.

| Item                     | Value |
| ------------------------ | ----- |
| Clicks, 24 Aug to 20 Sep | 89    |
| Clicks, 21 Sep to 18 Oct | 54    |

The homepage also became slower on mobile. In a local production-build Lighthouse test on 14 September, it scored 96 with a 1.9-second Largest Contentful Paint. On 20 October, the live site scored between 85 and 87 across three runs, with LCP between 3.0 and 3.2 seconds. We traced the regression to the branded intro cover not being skipped on phones as intended and an entrance animation on the hero headline. Both were fixed on 20 October. That fix is now deployed, but the scores reported here are the before-fix measurements. We need another comparable measurement to establish the result.

The stricter October review also left evidence and authority as open gaps. Those are current problems to handle, not proof that the September hygiene work failed.

## What I think the evidence says

Our audit score rose after we corrected measurable technical and content hygiene issues, then received a lower score under a stricter independent review. The score is a diagnostic summary, not a business outcome. Our organic clicks fell in the latest 28-day comparison. Bristol profile actions rose. National rankings and AI citations remain absent across the sampled terms. Each is a different measure, and rolling them into “SEO improved” would hide the differences.

The results have also changed how I think about our next work. We had spent time making a much larger set of URLs exist. The July analysis showed that many of those pages were not earning meaningful positions or clicks. We reduced that footprint and improved internal crawl paths. The next constraint is not another batch of city pages or another round of markup. It is whether the pages we want to rank have enough distinct value, proof and outside authority to deserve that visibility.

## What we are doing next

First, we are measuring the homepage again after the mobile intro and animation fixes, using the same local Lighthouse approach. The October numbers above are the pre-fix baseline, so I will not claim a performance recovery until we have a comparable post-fix result.

Second, we are getting our two unindexed service pages, [SEO in Bristol](https://aristral.com/bristol/seo-services) and [SEO in Derby](https://aristral.com/derby/seo-services), recrawled, with stronger internal links pointing at them.

Third, we are prioritising evidence. We need dated, permissioned case studies with baselines and outcomes before making stronger claims about client results. We also need to earn links and mentions from relevant sites, so that other organisations can independently vouch for our work.

Finally, we are keeping the cleanup set under observation until Google has dropped those noindexed URLs. We will review clicks, impressions, indexation and local profile actions as separate measures. If they move in different directions, the report should say so.

The next job is to improve the evidence and authority behind the pages that matter, then measure again.

## Methodology

Every Aristral article is researched and written by the practitioners who build the systems they describe. We verify each statistic against its primary source at the time of writing, link out to that source, and date the claim. SERP and keyword figures are pulled from live commercial APIs rather than estimated. We don't accept affiliate revenue or sponsorship from any tool or vendor named in our content. Spotted something out of date? Email [admin@aristral.com](mailto:admin@aristral.com) and we'll correct it.

About the author

### Taha Bilal

Co-founder, Aristral

Taha Bilal is a co-founder of [Aristral](https://aristral.com/), a UK AI automation and SEO agency based in Clifton, Bristol. He has been running SEO and digital-growth campaigns for SMB and SaaS clients since 2018, and now leads Aristral's combined SEO + GEO programmes for service businesses across the UK and US. Corrections and source requests: [admin@aristral.com](mailto:admin@aristral.com).

[LinkedIn](https://www.linkedin.com/in/tahabilal366/)

Want this built for you?

Book a free 30-minute call and we'll map the fastest path for your business.

[Book a 30-minute call](https://aristral.com/book)

Filed under: [SEO](https://aristral.com/blog/category/seo)

## Related articles

1. ### [How to measure AI search visibility](https://aristral.com/blog/ai-visibility)  
25 Sep 2026 SEO 9 min read
2. ### [llms.txt: what it is, who uses it and whether it helps](https://aristral.com/blog/llms-txt)  
25 Sep 2026 SEO 8 min read
3. ### [Automated SEO reporting: what we automate across 12 sites](https://aristral.com/blog/automated-seo-reporting)  
25 Sep 2026 SEO 10 min read

---

Markdown version of this page. Human page: https://aristral.com/blog/90-day-audit-what-moved-our-rankings-and-what-didnt-transparency-post. Site index for agents: https://aristral.com/llms.txt. Contact: https://aristral.com/contact