Discussion
  • SEO Analytics

An algorithm update and your own release land in the same week. How do you tell which one moved traffic?

Published 1 reply

Discussion starter: an SEO measurement question, posted with an answer from Niraj Raut to open the thread. If you have dealt with this on a site, reply with what you saw, especially where it differs.

Disclosure: Niraj Raut, who posted this answer, runs the SEO consultancy linked at the end of it.

1 reply

  • admin
    Admin
    Original poster

    The sitewide traffic line cannot separate them. Their footprints can. A release leaves a pattern that follows your code: particular templates, URL patterns, devices or locales, a known deploy time, and, if something broke, a trace in tracking, crawl or indexing data. A Google update follows Google’s systems: it cuts across your templates, shows up in queries where competitors move too, affects Google only, and lands inside the rollout window. When the footprints overlap, reverting part of the release on a subset of pages gives stronger evidence than any amount of chart reading.

    The order of work matters. Google’s core updates documentation recommends waiting at least a full week after a core update completes before analysing your site in Search Console, then comparing that week with a week before the rollout began. A release that broke something should not wait that long, so check the release first and assess the update on Google’s timetable.

    Step 1: confirm traffic moved, not tracking

    Releases break measurement too: a removed tag, a consent banner change, a redirect that strips the referrer, or a single-page app route that stops firing page views. Compare GA4 organic sessions with Search Console clicks for the same days. If GA4 fell and Search Console clicks did not, you are looking at a measurement fault, most likely introduced by the release, not a ranking change. If both fell, the traffic loss is real.

    Use clicks as the main series rather than impressions. Impressions can shift for reasons unrelated to either cause: when Google stopped supporting the &num=100 results parameter in September 2025, many sites saw desktop impressions fall and average position improve, while clicks were affected far less, according to Brodie Clark’s analysis. Google’s guide to debugging search traffic drops also points to Search Console’s data anomalies page for known reporting issues, so check it for the week in question.

    Step 2: pin down the timing

    • Collect real deploy times, including CMS edits, CDN or edge rules, feature-flag changes and configuration changes that bypass the main release process.
    • Take the update’s start and end dates from the Google Search Status Dashboard, as the core updates documentation advises.
    • Read daily data first. For the most recent days, the Search Analytics API has supported hourly data since April 2025, which can show whether a break happened at a particular hour. Pull it while the week is fresh.
    • Add both events to the Performance charts with Search Console custom annotations so everyone reading them sees the same timeline.

    Interpret the timing with care. A tracking break appears within hours of a deploy. A release that changes what Google indexes or how pages rank appears gradually as pages are recrawled, which can look like an update rolling out. So a step at the deploy hour points strongly at the release, but a gradual slide does not clear it. The absence of an announcement does not clear Google either: the same documentation says Google makes smaller core updates that “are not announced because they aren’t widely noticeable”.

    Step 3: compare what the release touched with what it didn’t

    List the templates, URL patterns, devices and locales the release changed, and compare their change in clicks with the change on everything else over the same days.

    • Touched and untouched segments fell by similar proportions: the release is unlikely to be the main cause.
    • Touched segments fell and untouched ones held: the release is the leading suspect, even with an update rolling out.
    • Both fell, touched segments harder: you probably have both. The untouched segments give a rough estimate of the update effect, and the extra drop on touched segments is the release’s likely share.

    Watch for hidden reach. Global navigation, internal link modules, robots.txt, sitewide JavaScript, rendering changes and CDN rules can touch every page even when the ticket mentions one template. If the release had sitewide reach, this comparison loses its control group and the later steps carry more weight.

    Also cut by query type. Aleyda Solis’s analysis of the May 2026 core update found that intent, market fit and source type drove the biggest visibility shifts, which is a different axis from your templates. If the loss clusters in one intent class, for example informational queries across several templates, that fits an update better than a release.

    Hypothetical example:

    An ecommerce site ships new product page templates on a Tuesday morning; category pages are untouched. A core update starts rolling out the next day. Two weeks later organic clicks are down 14%: product pages are down 20% and category pages 8%. Bing traffic to product pages has fallen too, and Page indexing shows a jump in “Alternate page with proper canonical tag” on product URLs, traced to a canonical tag that now points every variant at its parent product. The category pages suggest the update cost roughly 8%; most of the extra loss on product pages belongs to the release, and the canonical bug gets fixed that day rather than after the rollout.

    Step 4: use controls the update cannot touch

    Bing. A Google update does not change Bing’s rankings. If Bing organic traffic to the touched templates fell at the same time, suspect the release. Two caveats: Bing crawls on its own schedule, so a release problem can surface there days apart from Google, and on sites where Bing sends little traffic the series may be too noisy to read.

    Competitors. Google’s core updates documentation describes them as broad changes that don’t target specific sites or individual pages. If many sites in your results reshuffled over the same days, that fits the update. If your pages fell while competitors mostly held their positions, apart from filling the gap you left, look harder at the release.

    Step 5: look for the release’s fingerprints in crawl and index data

    • Crawl stats: changes after the deploy in response codes, host status, average response time or file types requested.
    • Page indexing: rises in “Excluded by ‘noindex’ tag”, “Alternate page with proper canonical tag”, “Duplicate, Google chose different canonical than user”, “Soft 404”, “Page with redirect” or “Crawled - currently not indexed” on touched templates.
    • URL Inspection: on a sample of touched URLs, the indexed result for the Google-selected canonical, and a live test for the rendered HTML, the declared canonical and indexability. The live test cannot show Google’s canonical choice, which is made at indexing time.
    • Server logs: Googlebot requests by template before and after the deploy, and the status codes it received.

    An update does not change the status codes your server returns or the canonical tags in your HTML. If these reports moved on touched templates, the release has a problem to fix regardless of what the update did.

    Evidence Points towards the release Points towards the update Where to check
    Tracking GA4 fell while Search Console clicks held (a measurement fault) Both fell together, which rules out tracking but not the release GA4 against the Performance report
    Timing A step at the deploy hour Movement inside the Status Dashboard window Hourly API data, daily Performance data
    Scope Concentrated on touched templates, devices or locales Spread across templates, clustered by intent Performance report filtered by URL pattern and query regex
    Bing Bing fell on the same pages Bing held Bing Webmaster Tools, analytics
    Competitors Your pages fell; others mostly held Many sites in the same results reshuffled Rank tracker with competitor sets
    Crawl and index Status code, canonical, noindex or rendering changes on touched templates No crawl or index anomalies Crawl stats, Page indexing, URL Inspection, server logs
    Partial rollback Reverted pages recover relative to the rest Reverted and unreverted pages move together Performance report by URL group

    Step 6: if it is still ambiguous, revert part of the release

    If the evidence still splits, revert the release, or the component you suspect, on a random subset of the touched pages and leave the rest as shipped. If the reverted subset recovers relative to the rest after Google recrawls it, the release was a cause. Requesting indexing through URL Inspection for a handful of representative URLs can prompt the first recrawls; the rest follow Google’s normal crawl schedule. Releases built behind feature flags make this cheap. Releases that were not can make it expensive, which is a reason to build the next one differently.

    When the answer is both

    A release can change how your pages fare under an update’s systems. A new template that thins out product copy, for example, might only cost rankings once the update re-evaluates the site. That interaction does not split neatly into two shares. Report what the evidence supports: the share you can attribute with a control (touched versus untouched, reverted versus not), and the remainder as consistent with the update.

    Make the next overlap easier to read

    • Ship SEO-relevant changes behind flags, section by section, so there is always an untouched comparison group.
    • Keep a release log with timestamps and the URL patterns each change touched.
    • Annotate deploys in Search Console on the day they ship.
    • Crawl the release candidate and production with the same settings, and diff titles, canonicals, status codes, robots directives and internal link counts before launch.
    • If an update is announced during a planned launch window and the business can wait, hold template-wide changes until the rollout completes.

    If I had to decide with incomplete evidence, I would let the segment comparison and the Bing check lead. If touched and untouched segments fell by similar amounts, Bing held and the crawl and index reports are clean, I would stop auditing the release and treat the loss as an update question. If any of those checks points at the release, I would start the partial rollback straight away rather than wait for the rollout to finish.

    Need help with this on your own site?

    Niraj Raut works with businesses in Nepal, Australia, the UK, Europe and the US on technical SEO, ecommerce SEO, local SEO and AI search.

    See SEO services

Add a reply

Replies are open to approved members. Log in or apply to join.