What Changes on Client Websites: 5 Months of Monitoring Data
How often do websites change in ways that matter for SEO and tracking? On the 60 client websites monitored in Deltio between April and September 2026, the answer is: more often than a monthly check would catch. A quarter of the sites had at least one robots.txt change, half had sitemap changes, and 10 domains are currently within 60 days of expiry.
This is a small study. The numbers below come from Deltio's own anonymised monitoring data, counted in aggregate. No site, client or domain is identified, and nothing here should be read as a statement about the web in general.
Method and limitations of this website changes data
The sample is every site monitored in Deltio as of 13 September 2026: 60 sites across 12 accounts. It is heavily concentrated. Three agency accounts hold 24, 14 and 13 sites, which is 51 of the 60 (85%). The other nine accounts have one site each. Most sites are European agency clients, so the mix reflects the kind of businesses those agencies work for, not a random slice of the internet.
Sites were added from 20 April 2026 onwards, so not every site was monitored for the full period. Each metric also has its own window, because checks were introduced at different times. Server availability checks, for example, only started on 2 July 2026. Where a window matters, it is stated in the section.
Finally, a recorded change is not the same as a mistake. Many of the events below were intentional edits. The data tells you that something changed, not whether it should have.
Robots.txt changed on 15 of 60 sites
Between 19 May and 9 September 2026, Deltio recorded 19 robots.txt events on 15 sites, which is 25% of the sample. Most were edits to the file's content: 13 events on 11 sites. The rest were more abrupt. On 3 sites the file disappeared, and on 3 sites a robots.txt appeared where there had been none.
A robots.txt file is short and rarely looked at, which is exactly why changes to it go unnoticed. A single Disallow: / line pushed from a staging environment can block crawling of an entire site, and a file that vanishes removes whatever rules the SEO team had put there on purpose.
For an agency, the practical step is to keep a known good copy of each client's robots.txt and compare against it whenever the file changes. Our article on robots.txt change alerts covers which kinds of edits deserve an immediate look.
Sitemaps are the busiest signal by far
Sitemap changes dwarf everything else. From 22 April to 13 September 2026, Deltio logged 81,964 sitemap URL changes on 30 sites, half of the sample.
The split is lopsided. Modified URLs (a new lastmod or changed content) account for 68,579 events on 28 sites, roughly 84% of the total. Added URLs came to 6,182 events on 24 sites. Removed URLs came to 7,203 events on 16 sites.
Two things stand out. First, removals outnumbered additions across the sample. Second, the modified count is inflated by how some CMSs and plugins handle lastmod, since a single bulk update can touch every URL in a sitemap at once. That makes raw modification counts a poor alert trigger on their own.
Removals are the part worth watching. A URL leaving the sitemap can be a planned cleanup, or it can be a category page that stopped being published after a plugin update. Reviewing removals per site, rather than the whole firehose of changes, keeps the signal manageable. The pages removed from sitemap guide explains how to tell the two apart.
On page SEO changes clustered on a few sites
During the monitoring period, Deltio recorded 121 on page SEO changes on 9 sites (15% of the sample). By element:
| Element | Events | Sites |
|---|---|---|
| Title | 41 | 6 |
| H1 | 34 | 6 |
| Meta description | 28 | 5 |
| Canonical | 18 | 2 |
The pattern is concentration. 121 events on 9 sites means these changes were not spread evenly; a small number of sites were being actively edited while most stayed stable. Canonical changes were the rarest by site count, but 18 events on only 2 sites is a lot of canonical movement for two properties.
No noindex events were recorded in this dataset. The check ran, and it found nothing to report in this sample.
What to do with this: on sites that are being actively worked on, title and canonical changes on pages that already rank deserve a second look by whoever owns SEO for that client.
Tracking and consent changes were rare
Only 1 site lost its GA4 tag, with 2 disappearance events, both on 16 June 2026. Consent platforms changed on 2 sites, for 8 events in total between 31 May and 8 July 2026. On both of those sites, the cookie banner or consent platform was at some point no longer detected (4 events). The remaining events were changes involving Cookiebot (3 events, 1 site) and CookieYes (1 event, 1 site).
Rare does not mean low impact. A missing GA4 tag leaves a gap in reporting that cannot be backfilled, and a consent banner that stops loading also raises a compliance question for a European business.
Availability: 1.7% of checks failed
From 2 July to 13 September 2026, Deltio ran 25,463 availability checks. 25,008 came back online (about 98.2%), 438 were offline (about 1.7%), and 17 were blocked by a firewall (under 0.1%). Failed checks were concentrated: 7 sites, about 12% of the sample, had at least one.
No site was in maintenance mode during checks, and no SSL certificate was within 14 days of expiry or already expired. Firewall blocks are worth separating from real downtime, because a security rule that blocks a monitoring service may also be treating some crawlers with suspicion. Whether short outages affect rankings depends on duration and timing, which the downtime and SEO article goes through.
10 domains expire within 60 days
As of 13 September 2026, 10 of the 60 monitored domains expire within 60 days, about one in six. This is a snapshot of upcoming renewal dates, and some of those domains may well be set to auto renew. The risk is the ones that are not, or where the card on file has changed. A quick confirmation with each client, well before the date, is cheaper than recovering a lapsed domain. See domain expiry monitoring for what to check.
Summary of the findings
| Signal | Window | Events | Sites affected (of 60) |
|---|---|---|---|
| Robots.txt changes | 19 May to 9 Sep | 19 | 15 (25%) |
| Sitemap URL changes | 22 Apr to 13 Sep | 81,964 | 30 (50%) |
| On page SEO changes | Monitoring period | 121 | 9 (15%) |
| GA4 tag disappeared | 16 Jun | 2 | 1 |
| Consent platform changes | 31 May to 8 Jul | 8 | 2 |
| Failed availability checks | 2 Jul to 13 Sep | 455 of 25,463 | 7 (about 12%) |
| Domains expiring within 60 days | Snapshot, 13 Sep | n/a | 10 (about 17%) |
What this data does not show
It does not show how often websites change in general. 60 sites, mostly from three agency portfolios, cannot stand in for anything larger, and one busy portfolio can move a number on its own.
It does not show intent or harm. A title change may have been a planned improvement, and a removed sitemap URL may have been deliberate. None of the counts measure traffic or ranking impact.
It does not compare periods fairly. Sites joined at different dates and metrics started at different times, so totals are not directly comparable across rows.
What it does suggest, for this sample, is that changes happen on client sites often enough that checking them by hand once a month would miss most of them. If you want the same view on your own client list, you can add your sites to Deltio.
Frequently asked questions
- How was the data anonymised?
- Only aggregate counts were extracted: totals of events and the number of distinct sites affected per change type. No URLs, domains, page content or account names were part of the export, and nothing in the article can be traced back to a specific site or agency.
- Why do the metrics cover different date ranges?
- Sites were added to Deltio gradually from 20 April 2026, and some checks were introduced later than others. Availability checks, for instance, began on 2 July 2026. Each figure is reported with its own window instead of forcing everything into one shared period, which would have meant discarding data.
- Why is the sitemap number so much larger than everything else?
- Each sitemap event is counted per URL, while robots.txt or title changes are counted per file or page element. A site with thousands of URLs that refreshes lastmod values in bulk can generate thousands of events in a single day, so the sitemap total reflects site size and CMS behaviour as much as editorial activity.
- Will this study be updated?
- The plan is to rerun the same counts as the sample grows, using the same definitions, so results can be compared over time. A larger and less concentrated set of sites would make the percentages more stable than they are with 60 sites across 12 accounts.