If your traffic chart shows a cliff sometime after late March, you're not imagining it. Google's March 2026 core update triggered ranking declines on roughly 71% of monitored affiliate domains, and the May 2026 update that followed only confirmed the direction — it rewarded the same signals of firsthand experience and original data. This isn't a story about affiliate sites getting singled out. It's a story about what "content" means to Google now, and it applies to product blogs, comparison pages, and SaaS content just as much as review sites.
Key Takeaways
- Affiliate sites saw the highest negative-impact rate of any content category — 71% experienced ranking drops, with typical losses of 20-35% and some pages losing over 50% (Digital Applied, 2026).
- AI-generated content published without human editorial oversight lost 60-80% of its traffic (Launchcodex, 2026).
- Sites publishing original research saw an average 22% visibility boost across a tracked sample of B2B SaaS sites (BuildMVPFast, 2026).
- Images under 1,200 pixels wide are now functionally invisible in Google Discover, regardless of how good the writing is.
What Actually Happened in the March and May 2026 Core Updates?
Google shipped three major quality-signal changes in quick succession: a spam update on March 24 that targeted manipulative link schemes, a core update on March 27 that re-evaluated rankings with those signals stripped out, and a May 2026 core update that doubled down on the same direction. The sequencing wasn't a coincidence. The spam update targeted manipulative link schemes and scaled content abuse first, then the core update immediately re-evaluated quality rankings with those signals removed.
For sites that had been leaning on thin, reworded, or AI-assisted content, the two updates compounded. Early data from multiple tracking tools showed affected sites losing 20-35% of traffic in the first week, with some domains losing over half their traffic on their worst-hit pages. This wasn't a narrow correction — the update affected an estimated 40-60% of websites overall, with affiliate sites the hardest hit at 71%.
Google's own language for what it's penalizing is specific: scaled content abuse. That's a deliberate phrase, and it's worth sitting with — it isn't about AI tools themselves, it's about volume without value.
Why Did Affiliate and Summary Content Get Hit So Hard?
Because most of it does the same thing search itself now does automatically: restate what other pages already said. The update didn't target affiliate marketing as a category — it targeted affiliate content that doesn't do anything the brand's own site doesn't already do better. Three formats absorbed most of the damage.
AI-assisted reviews with no human testing. Sites running AI-generated reviews with no original opinion, no hands-on experience, and nothing the manufacturer couldn't have written themselves saw the biggest drops. One affiliate program manager described cutting ties with 18 publishers in a single week after the update, because their content read like it had zero human involvement — and their referral traffic had already dried up before the calls were even made.
Templated comparison pages. Generic "Product A vs. Product B" tables built from spec sheets rather than actual testing lost ground across the board, especially where every competitor's page contained the same information in a different order.
Programmatic and coupon-aggregator content. Thin, affiliate, and templated content was hit hardest of any category, with AI content farms losing 60-80% of their traffic. Coupon sites relying on dynamically generated codes with no editorial layer saw pages systematically de-indexed rather than merely demoted.
The common thread across all three: the content existed to occupy a keyword, not to contain information a reader couldn't get somewhere else.
What Actually Gained Ground?
If the losers were generic, the winners were specific. Analysis of over 2,000 domains found that first-party brands and government sites were the biggest gainers — IMDB, Amazon, and Apple among them — while aggregators, user-generated content platforms, and thin affiliate sites lost ground. Original sources beat sources about sources.
That pattern held at the content level too, not just the domain level. A tracked sample of 32 B2B SaaS sites that published original research saw an average 22% visibility boost, and the top three results for many queries shifted toward authoritative brand domains rather than third-party commentary. Meanwhile, even large, established publishers weren't exempt from the correction if their content stayed generic — one major SaaS blog lost 70-80% of its organic traffic on broad, top-of-funnel content that competed on volume rather than depth.
The practical read: authority isn't inherited from domain age or backlink count anymore. It's earned per-page, by whether that specific page contains something no other page does.
The Image Rule Nobody's Talking About Enough
Buried inside the broader quality shift is a technical requirement that's easy to miss and expensive to ignore: image resolution. Following a Discover-specific core update in February 2026, images now need to be at least 1,200 pixels wide to qualify for large thumbnails, ideally in a 16:9 aspect ratio matching how content cards render in the feed, with the max-image-preview:large meta tag enabled to actually display them at that size.
This isn't a minor polish item. Posts with 1,200px+ images and the correct meta tag saw 45% higher click-through rates in Discover compared to pages with smaller or missing images, and sites below that threshold are effectively excluded from large-format placement no matter how strong the writing is. Some sites reported partial Discover recovery within 7-10 days simply from upgrading their hero images — it's one of the fastest technical fixes available, alongside removing clickbait framing from headlines.
If you're publishing well-researched, original content and still losing Discover traffic, check your image dimensions before you touch anything else.
Does Your eCommerce Platform Choice Affect Any of This?
Indirectly, yes. The core updates reward pages that can carry original product data, fast-loading images at the right resolution, and genuine first-party detail — and some platforms make that easier to execute at scale than others.
Shopify remains the default for teams that want managed hosting and a large app ecosystem, though customizing templates to add the kind of original testing content Google now rewards often means working around theme constraints. WooCommerce gives full control since it's WordPress-based, which suits sites that already run a content-heavy blog alongside the store, at the cost of needing to manage hosting and plugins yourself. BigCommerce sits in between — strong native features, less flexible on deep customization.
OpenCart is worth a specific mention here because it's open-source and free to self-host, which matters if you're trying to run original data collection (pricing tests, inventory experiments, genuine product comparisons) without per-transaction platform fees eating into the margin that funds that research. If you're evaluating options and want to build an eCommerce store free, OpenCart is one of the few platforms where that's literally true rather than a free-trial funnel into a paid tier.
None of this replaces the content and image fixes above — a well-optimized platform still won't rank a page that just restates the top three competitors.
A 10-Minute Self-Audit for Your Existing Content
You don't need an enterprise SEO tool to know whether a page is at risk. Run these three checks on your highest-traffic pages:
- The rewrite test: Could a reader get everything on this page from the top three competing results, just phrased differently? If yes, it's summary content, no matter how well it's written.
- The intent test: Does the page serve the reason someone searched, or was it written primarily to contain the keyword? Google's own language treats these as separate signals now, not the same thing.
- The image test: Open your hero image in a new tab and check its actual pixel width. Anything under 1,200px is a Discover liability, independent of content quality.
Pages that fail the first two tests aren't necessarily worth deleting — but they need a genuine addition before the next core update: a test you actually ran, a number no competitor has published, or a screenshot proving you used the thing you're writing about.
Frequently Asked Questions
How long does recovery take after a core update penalty?
Recovery timelines vary by severity — minor fixes may show results within weeks, while deeper structural or content issues can take several months to stabilize, and Google tends to fully validate improvements during a subsequent core update cycle rather than continuously.
Does adding more content help recover lost rankings?
No — the update penalizes content published at high volume without meaningful original value, a pattern Google specifically calls "scaled content abuse". Adding more generic pages compounds the problem rather than fixing it.
Is AI-written content banned from ranking well?
Not inherently. The penalty falls on content published without substantive human editorial oversight — the tool used to draft it matters less than whether a person added real testing, opinion, or data on top of it.
Should I remove pages that lost rankings?
Not immediately. A better approach is pulling your top pages by pre-update traffic, comparing performance in Search Console, and prioritizing fixes on pages that were close to ranking well but lacked proof of real experience, rather than removing content wholesale.
How much do original images actually matter for ranking?
More than most sites assume. Large, high-resolution images increase click-through rates by up to 5% and are treated as a requirement for high-quality status in Discover specifically, separate from how the written content itself is evaluated.
