Core Updates and Affiliate Sites: How to Tell If You Were Hit, What to Change, and What to Leave Alone Until the Next One
Every few months Google reassesses how it ranks sites and some affiliate sites lose a third of their traffic in a fortnight. The owners who recover share one habit: they worked out whether it was the update at all, what the update measured, and changed that, then left the site alone until the next one.
The owners who don't recover rewrote titles the week of the drop, changed the theme, bumped every date, and were reassessed six months later as the same site with new titles.
This post covers how to date a drop against Google's update timeline and tell a site-wide hit from a page-level slide, the three content patterns affiliate sites get hit for, what to change in the window between updates and in what order, what to leave alone, and how to read the next update when it comes.
The not-ranking post covers the twelve reasons a site doesn't rank at all; this is the one reason that arrives on a date.
The short answer
Open Search Console, find the day clicks fell, and check it against Google's update timeline; if it lines up and most pages fell together, it's the update. Read the twenty pages that fell most for the pattern they share.
Fix the pattern across the site (verdicts, sameness, first-hand signal, pruning) starting with the pages that earned. Don't touch titles, dates, word counts or the theme. Expect the recovery at the next update, not next week.
Was it the update?
| Check | Core update | Something else |
|---|---|---|
| Date | The drop starts inside a published rollout window and holds after it ends | The drop starts on a date with no rollout, or recovers before the rollout ends |
| Breadth | Most pages fell, by a similar proportion, across topics | One page, one topic, or one keyword fell |
| Shape | Positions slid several places site-wide; impressions and clicks both down | One page vanished (technical or manual), or clicks fell with positions steady (a results-page change) |
| Competitors | Similar sites moved too, some down, some up | One competitor took one keyword |
Google publishes the start and end of each core update on its Search Status Dashboard, and a rollout runs two to four weeks. A drop that starts in the window isn't a verdict until the window closes; rankings move throughout and sometimes reverse. The week after the end date, compare the last 28 days with the 28 before, and that's the number.
The three patterns that get hit
Core updates measure the site, not the page, and on affiliate sites the site-level signal is a pattern repeated across guides. Three come up most:
- Sameness. Fifty guides with the same shape, the same seven picks, the same intro, and nothing in any of them that the Amazon listing didn't say. One such guide is thin; fifty are a site that adds nothing, which is what the update is built to find.
- No verdict. Guides that list ten products and end with "it depends on your needs". A reader who got no answer is a signal, and a site full of them is a pattern. The verdict box post covers the fix per page.
- No first-hand signal. No method, no author, no evidence anyone chose these products for a reason. The E-E-A-T post covers what counts on a site working from listing data; the update checks whether any of it is there.
To find yours, take the twenty pages that fell most and read them as a stranger, in a row. The pattern is usually visible by the fifth.
What to change, in order
- Prune first. Pages with no traffic, no links and no prospect: delete or merge. A smaller site with a better average is reassessed better than a bigger one dragging the same dead weight.
- The earners next. The twenty pages that earned before the drop, rewritten against the pattern: a verdict with a reason, a method line, a section the listing couldn't have written, picks checked against today's catalogue.
- Then the hubs, so the rewritten guides sit in a structure that says the site knows its category.
- Then the rest, at a pace you can sustain until the next update, because the next update is the deadline.
- Keep a log of what changed on which page and when. When the next update lands, it's how you tell which change worked.
What to leave alone
- Titles. Rewriting fifty titles the week of a drop resets what those pages had and changes nothing the update measured.
- Dates. Bumping "last updated" on unchanged pages is the opposite of a fix.
- Word counts. Longer isn't the signal; the update doesn't count words, it reads them.
- The theme, the URL structure, the hosting. None of them caused a site-wide content reassessment and all of them add noise to the next one.
- Disavowing links. A core update isn't a link penalty; a disavow file does nothing here and can hurt.
- Daily rank checks. The number to watch is the 28-day comparison after the next rollout ends; everything in between is weather.
Reading the next update
Recovery, when it comes, usually comes at a core update, because that's when the site is reassessed. Some sites recover partway at an intermediate update; most wait for a core one. When the next rollout is marked complete, compare 28 days against the previous 28, by page, against the change log.
Pages that were rewritten and rose confirm the pattern you fixed; pages that were rewritten and didn't suggest the pattern was something else, and the twenty-page read starts again. A site that changed something real and got no movement at all is rarer than one that changed the wrong things.
Doing it with AffSERP
The patterns core updates hit are the ones AffSERP's X Best Review tool is built to avoid: each guide gets a verdict with a reason, a how-we-chose section, and picks selected from current product data rather than the same seven every time, so fifty guides don't share one shape.
For a site already hit, the Refresh option regenerates the earners on the same URL with current picks and the verdict in place, which is step two of the order above run as one action; the pruning, the hubs and the log are yours. Nothing in the tools touches titles or dates for the sake of it.
FAQ
How do I know if my site was hit by a Google core update?
Three checks. Date it: open Search Console's Performance report, find the day clicks fell, and compare it with Google's published update timeline; a drop that starts within the rollout window and holds after it is the update. Measure the breadth: if most pages fell together it's site-wide, which is what a core update does; if one page fell it's that page. Rule out the rest: a new competitor on one keyword, a lost pick, or a technical fault fall on one page or one date, not across the site during a rollout.
Can an affiliate site recover from a core update?
Yes, and usually not until the next core update or a later one, because a core update reassesses the whole site and the reassessment is what lifts it back. Sites that recover changed something real in the window between: thin guides merged or rewritten with verdicts, sameness broken up, first-hand signal added, dead pages pruned. Sites that didn't recover either changed nothing or changed the wrong things (titles, dates, word counts) and were reassessed as the same site.
What should I change after a core update traffic drop?
The pattern, not the pages. Pick the twenty pages that fell most and read them as a stranger: do they share a shape (same ten picks, no verdict, nothing the listing didn't say)? Fix that shape across the site, starting with the pages that used to earn, and prune what isn't worth fixing. Don't rewrite titles, bump dates, add words or swap themes; none of those change what the update measured, and a flurry of them resets the pages' history for nothing.
How long does a core update take to roll out?
Google's recent core updates have rolled out over two to four weeks, with Google posting the start and end dates. Rankings move throughout the window and can reverse partway, so a drop on day three isn't a result until the rollout is marked complete. Read the Performance report the week after the end date, compare 28 days against the previous 28, and treat that as the verdict.
AffSERP's X Best Review gives each guide a verdict with a reason, a how-we-chose section and picks from current data, and Refresh rewrites the earners on the same URL.
Start with free wordsKeep reading
