What the March 2026 Core Update Taught Us About Programmatic SEO

How do you recover from a core update when your site is built on hundreds of thousands of programmatic pages? You stop asking Google to index pages that give it nothing. You publish content only you could write. And you connect your pages with internal links so your strongest content doesn't sit in a silo.

That's the short version. Here's the long one, with data.

The setup

MartiniWorks helps people build their cars the right way, and they sell aftermarket wheels, tires, and car parts. The platform we built for them generates hundreds of thousands of pages programmatically across their store, one for every combination of vehicle and part a shopper might search.

This is how you built ecommerce SEO before 2024. Cast the widest possible net and let the long tail pay for itself. The pages that matched a real search paid for the ones that didn't, and the ones that didn't cost nothing.

Then Google started to crack down on this. With its March 2024 core update, Google began scoring quality at the site level and named scaled content abuse as a spam policy. Thin pages stopped being free. Every URL you ask Google to crawl became part of how it judged the rest of the domain, whether it made the index or not.

What we saw is that fewer than 10% of those pages ever earned an impression. Under the old rules, that was harmless. Under the new rules, it was a liability waiting for an update to bring down our overall search presence.

For MartiniWorks, March 2026 was that update.

What March 2026 changed

Google shipped two updates within 72 hours of each other. The March spam update ran on March 24 and finished in under 20 hours, the fastest confirmed spam rollout in the Search Status Dashboard's history. The March core update followed on March 27 and rolled out over 12 days.

A core update is a re-rank of everything at once. Google publishes the dates, never the reasoning. The explanations you read afterward, page-level scoring, information gain, a crackdown on scaled content, are all inferred from watching who moved.

So we skipped the theorizing and watched our own pages.

The updates didn't move the domain as one block. They moved it by page type. Pages built on original content and perspectives climbed. Pages that restate information available anywhere else fell, hard, no matter how accurate they were.

Whatever the mechanism, Google was discounting content produced at volume without editorial judgment behind it.

That describes most programmatic SEO. Including ours.

The dip

MartiniWorks came into March at its strongest organic baseline to date. By mid-May, daily organic clicks were down nearly 25%.

A quarter of organic traffic, gone in six weeks.

The instinct in that moment is to publish more. More pages, more keywords, more surface area. The data said to do the opposite.

So we made three changes.

Gate the index

We stopped treating "the page is dynamically built and may eventually have content on it" as a reason for the page to be indexed.

We built an indexability gate into the platform. It watches what every programmatic page actually has to offer, and a page with nothing to show gets noindexed, removed from the sitemap, removed from links across the site, and ultimately taken out of Google's consideration. A page that gets content in the future re-enters the index on its own.

The premise is simple. Google had already passed judgment on these pages. Most sat in "crawled, currently not indexed" and never earned an impression. We stopped asking it to reconsider.

That reduces crawl waste. It concentrates quality signals on pages that deserve them. It tells Google the site knows the difference.

Publish what only you can publish

While the indexability gate was being built, we pulled the numbers on which pages survived or even thrived during the update.

Pages with original editorial content, backed by our own recommendations, data, and reasoning, saw a 143% increase in organic traffic. MartiniWorks also captures and presents first-hand community content. Those pages saw a smaller increase of 17%, but an increase nevertheless.

On the opposite side, templated informational pages that tended to restate already existing information dropped 41%. Before the update, those templated pages were the strongest performers on the domain.

The pattern was hard to argue with.

A brand's history is the clean example. It's accurate, it's relevant, and it's on forty other websites. After March, that kind of page stopped performing.

Our theory: Google is cracking down on recycled content to fight the flood of AI-generated pages. The counter is content with perspective, opinions, and conclusions.

That kind of content pays off twice. It's also what AI engines cite when they answer a question in your space, the same dynamic we covered in how to make your brand show up in AI search.

So the content push changed shape. Guides with real recommendations and the reasoning behind them. First-hand content from actual owners, with costs, failures, and photos. Conclusions drawn from our own sales data. Content that requires someone to have an opinion and sign their name to it.

Internal links, the old-fashioned tactic

The third change is as old as SEO itself. Internal links pass link equity. A link from one page hands a share of that page's authority to the page it points to.

Your site earns authority from the outside. Internal links decide where it goes.

We audited every landing page template on the site. Fourteen of fifteen had zero links to the original content that was winning.

The original content Google was actively rewarding sat in its own silo while hundreds of thousands of templated pages linked only to each other.

So we wired the site together with intention. Content pages now link to the money pages they naturally support, and money pages link back into the content that's relevant to them. The content earns the authority. The links route it to the pages that convert.

It's a UX improvement as much as an SEO one. A reader on a content page gets a direct path to buy. A shopper on a money page gets proof and context one click away. Google rewards pages that are valuable to real human beings, not just to search engines, and a link a person would actually click is exactly that.

Nothing clever. Just equity flowing where it hadn't been.

The results

The May core update rolled out from May 21 to June 2. The site that had lost a quarter of its traffic in March held steady through it, and daily clicks ticked up during the rollout window itself.

The shape of the traffic tells the better story. Impressions fell about 40% from the March peak, and that was Google's doing, not our pruning. The pages we removed had no impressions to lose in the first place. Google decided what to stop showing. What we controlled was the site it would re-evaluate next time.

Clicks stabilized while impressions kept falling. Sitewide click-through rate climbed more than 20%.

Fewer pages. Fewer keywords. The same clicks from a far smaller, far cleaner footprint.

Then came the June 24 spam update, and the page types started moving the right direction. Original editorial content now earns 56% more daily clicks than it did before the March update. The templated pages, the type that fell 41%, grew roughly 35% in the weeks after the June rollout. Those pages now carry internal links to the original content that vouches for them.

The pages we invested in are growing, and the pages we pruned are no longer dragging them down.

What we learned

Programmatically generating pages to catch keyword searches, on its own, is no longer a strategy. In a competitive market it's a liability. Volume without substance is exactly the thing Google has built to catch over the past few years, and it shows.

Programmatic pages are not necessarily the wrong move. They just need to be intentional and provide value to real humans. That means gating what gets indexed, so every URL you hand Google has something on it, and building original content around the programmatic core instead of shipping the core alone. Then link the two together. Your best content can't lift the rest of the site from inside a silo.

And first-party perspective is the moat. The internet does not need another page of recycled information. Google and AI models are both learning to filter that out. What they can't filter out is data only you hold, conclusions only you can draw, and experience only your customers have had. That's what ranks in search, and it's what earns the citation in an AI answer.

The web has plenty of information. What ranks now is knowledge.

FAQs