Shopper behavior is changing. Are SEO teams chasing outdated KPIs?

Clicks are down.

Revenue is up.

If you work in SEO for an ecommerce site, that probably sounds familiar. It might also sound annoying, because the old reporting model trained us to expect a much cleaner relationship between traffic and revenue.

More rankings meant more clicks. More clicks meant more sales. If organic traffic was growing, SEO was working. If traffic was falling, something was wrong.

That story is getting harder to defend.

Across SearchPilot's ecommerce customers, we have been seeing a pattern that does not fit the old dashboard. Year on year traffic is down, but revenue is stable or rising. That made us ask a simple question: where are the clicks going?

Or, more practically: how do you prove SEO still matters when one of the most familiar SEO KPIs looks worse than the business result?

That was the focus of this webinar. I wanted to talk about what is happening to the purchase funnel, why some old KPIs are becoming less useful, which newer signals deserve more attention, and why testing matters even more when the dashboard starts telling a confusing story.

Key takeaways

• Clicks can fall while revenue rises because more research is happening before the user reaches your site, inside SERPs, product grids, AI Overviews, and LLM chats.
• SEO has always had a Red Queen problem: teams have to keep moving to hold their position. The difference now is that total clicks compared to your performance last year are no longer the metric you need to tell whether you are moving fast enough.
• You may be winning even if traffic declines. Rankings can hold, product visibility can remain strong, and more valuable shoppers can still arrive later in the funnel.
• Year on year traffic, especially for PLPs, blog content, and evergreen content, can now tell a misleading story.
• PDP clicks, PDP conversion rates, organic revenue, click through rate, and Merchant Center clicks and impressions can give ecommerce teams a clearer view of what SEO is contributing.
• PDPs are becoming more important as landing pages because shoppers are often closer to purchase by the time they click.
• SEO A/B testing with control buckets, and multi-metric testing matter more now because before and after reporting is too easily distorted by seasonality, algorithm updates, competitors, and SERP changes.

The old purchase journey had a lot more clicks

Think about how someone might have shopped online ten years ago.

They might start with a broad query like "best leather belts." That would send them to review sites, category pages, maybe some affiliate pages calling themselves review sites. Then they would refine the search. "Best leather belts for men under $100." More tabs. More pages. More clicking around.

At some point, they would land on an ecommerce site. Maybe a PLP. Maybe a blog post. Maybe a buying guide. From there they would compare options, click into a PDP, make a decision, buy, and maybe come back later.

That older funnel was extremely convenient for SEO reporting because much of the research journey happened on your site. Awareness, interest, consideration, intent, conversion, loyalty. You could see a lot of that path in analytics.

The user was messy, but at least the clicks were there.

The new purchase journey starts somewhere else

Now look at a modern ecommerce SERP, especially on mobile.

A shopper can see sponsored products, organic product grids, review snippets, prices, discounts, images, star ratings, filters, and an AI Overview before they ever reach your website.

That is a very different shopping experience.

In some cases, Google is starting to behave like the PLP. The shopper can compare products, check price ranges, scan reviews, and filter options without clicking through to a retailer. If they are using an LLM, they may ask a highly specific question and get a shortlist before visiting any site at all.

So awareness, interest, and consideration can all happen off your website.

That does not mean SEO has stopped mattering. It means the click may arrive later.

By the time someone reaches your site, they may already know what they want. They may not be browsing. They may not be comparing six tabs. They may be ready to buy.

That changes what the click means.

The gloomy future and the more likely one

There is a gloomy version of this future.

In that version, AI eats the whole funnel. The shopper asks ChatGPT, Google AI Mode, or another assistant what to buy. The assistant recommends products. Payments happen inside the assistant. The retailer gets the sale, maybe, but never gets the visit.

No PDP view. No email capture. No direct relationship. No site experience. No chance to build loyalty.

That is the "I drink your milkshake" version of ecommerce search.

I do not think that is the only possible future, or even the most likely near term version.

The more likely version is messier. AI and SERP features influence the research journey, but shoppers still want to click through when they are ready to buy. They still want to choose a product. They still want to land on a website they trust. They may use AI to narrow the options, then use Google to find the best retailer, or click from a product grid into a PDP.

In that version, your site still matters a lot.

But the job of the site changes. You may not be welcoming someone at the start of the journey anymore. You may be meeting them at the moment of intent.

That is a different kind of SEO opportunity.

Ecommerce is not shrinking

One reason I am not pessimistic about ecommerce SEO is that the underlying market is still healthy.

People are still buying online. The money is not vanishing. The journey is changing.

That distinction matters because it stops the conversation from becoming "SEO is dying" or "organic is over." I do not think that is what the data is saying.

The better read is this: some of the clicks we used to count as SEO success are being displaced. Earlier research clicks are being absorbed by SERP features, AI summaries, shopping modules, and LLM conversations. But the commercial demand is still there.

In some cases, the click that remains may be worth more.

That is uncomfortable for reporting, because it means a smaller traffic number can sit next to a stronger revenue number. But for the business, that is not necessarily bad news.

It does mean SEO teams need to stop telling the old story.

The Red Queen problem has changed

SEO has always had a Red Queen problem.

In Through the Looking Glass, the Red Queen has to keep running to stay in the same place. SEO has always felt a bit like that. You keep pages fresh. You improve templates. You fix technical issues. You update content. You respond to competitor moves. You deal with platform changes. You keep running.

For years, total organic clicks were one of the ways we judged whether we were running fast enough.

That signal is weaker now.

You may still be maintaining rankings. You may even be improving visibility. Your products may still be appearing in the moments that matter. But if the SERP now satisfies more of the research journey before the click, your total clicks can fall anyway.

That is the part that makes old dashboards dangerous. They can make a team look like it is losing when the reality is more complicated.

You can be doing good SEO work, holding ground in a harder environment, and still see year on year traffic decline.

That is a miserable sentence to put in a report, but it may be true.

Winning can look like less traffic

This is the uncomfortable bit.

Winning may now look like fewer clicks, but better clicks.

A shopper who lands on your PDP after seeing your product in a grid, reading an AI summary, comparing prices, and deciding they are interested is not the same as someone who clicked a top of funnel article out of curiosity.

The PDP visitor may be far more motivated.

That changes the value of the click. It also changes what we should measure.

If clicks are less common, each click may carry more intent. If users arrive later in the funnel, your PDP has to work harder. It may need to carry information that older site journeys assumed users would see earlier.

That means ecommerce teams need to stop thinking of PDPs as only the final page in the journey.

For many shoppers, the PDP may be the first page they see.

PDPs are becoming the new landing pages

Older ecommerce SEO models often assumed users would enter through PLPs, guides, category pages, or evergreen content. Those pages still matter. They can still feed search engines and AI systems with the information users need.

But the click may land somewhere else.

Product grids and AI-assisted journeys often surface PDPs more directly. If the product card has the image, price, merchant name, offer, review data, and availability, the user can choose a product before they ever touch your site.

When they do click, they may click straight into the PDP.

That means the PDP has to do more jobs.

It has to reassure the user that they are in the right place. It has to answer the last questions before purchase. It has to make trust signals obvious. It has to show delivery, returns, sizing, availability, reviews, specs, and any other detail that might remove friction.

If the user used to see that information on a guide or PLP before reaching the PDP, you may need to rethink the page.

The landing page is now the conversion page.

That is a real shift.

Stop overreacting to year on year traffic

The first KPI I would be careful with is year on year organic traffic.

I am not saying throw it away entirely. It still has uses. But if you are treating it as the main proof of SEO value, especially for ecommerce, it can mislead you.

This is especially true for PLPs and informational pages. Those page types often sit earlier in the old journey. If AI Overviews, shopping features, and LLM chats absorb more of the research stage, those pages may lose clicks even when their information is still influencing the purchase.

A guide may still matter if its content is being summarized or used to answer a shopper's question. A PLP may still matter if its structure, content, and product data help search systems understand your catalog. But the old click may not happen.

That makes the reporting harder. It does not make the work worthless.

The danger is that leadership sees a falling traffic chart and assumes SEO is failing. If revenue is rising at the same time, the SEO team needs a better explanation than "traffic is down."

The metrics I would watch more closely

The first metric I would watch more closely is clicks into PDPs.

That is where the money is likely to be. If users are arriving later in the purchase journey, PDP clicks tell you much more than broad traffic totals. They show whether motivated shoppers are reaching the pages where conversion happens.

Then I would look closely at PDP conversion rate.

If a user lands on a PDP ready to buy and does not convert, something may be missing. Maybe shipping is unclear. Maybe sizing is hard to understand. Maybe returns are buried. Maybe the page assumes the user already visited another page first.

That assumption may no longer hold.

Click through rate also matters, especially in ecommerce SERPs where many products and retailers appear side by side. If you are a third party seller, you might be competing with other retailers selling the same product. The question is not only whether your product appears. It is whether your listing earns the click.

Merchant Center clicks and impressions deserve much more attention too.

The product grid is not the same as the old ten blue links. Google Merchant Center data, feed quality, product titles, images, pricing, promotions, availability, and review signals can all affect how appealing your products look in shopping surfaces.

If you are not watching that data, you are missing a large part of the modern search experience.

And then, of course, revenue attributable to organic traffic matters. Make sure your analytics setup can show how organic users move through the site, whether they convert, and what those conversions are worth.

When leadership asks whether SEO is still working, revenue is the language that travels.

Some AI metrics are not ready yet

There are a few AI-related metrics I would keep an eye on, but I would not build the whole reporting story around them yet.

Brand sentiment in AI models is one. People are starting to ask how AI assistants talk about their brand. That is a good question, but the measurement is still early. Outputs vary. Personalization can change answers. The same prompt may not return the same response every time.

Share of voice in AI chats will probably become more useful as the tools mature and better analytics become available. But right now, AI query volume is still murky. We do not have the same kind of reliable search volume data for LLM chats that SEO teams are used to in classic search.

The same goes for query clusters inside AI tools. Interesting? Yes. Worth watching? Definitely. Ready to carry your board reporting? Not yet.

Do not ignore these areas. But do not pretend they are cleaner than they are.

Before and after reporting is too noisy

If you cannot rely on year on year traffic in the same way, how do you prove SEO is paying off?

This is where testing comes in.

A lot of teams still use before and after reporting. You make a change, wait, then compare performance before and after the change.

The problem is that everything else changes too.

Maybe you launched content right before a strong sales season. Maybe a competitor changed prices. Maybe a core update landed. Maybe a PR campaign drove attention. Maybe demand shifted. Maybe Google changed the SERP layout.

If performance went up, did your change help? Or were you going to have a good month anyway?

If performance went down, did your change hurt? Or did the market move under your feet?

Before and after analysis cannot reliably separate those things.

It has always been imperfect. Now, with shopper behavior changing so quickly, it is even more exposed.

SEO A/B testing becomes a tracking tool

SEO A/B testing gives you a cleaner way to measure impact.

Instead of comparing this year to last year, or the weeks before a launch to the weeks after, you compare a variant group against a control group in the same period. The control bucket helps account for seasonality, competitor movement, algorithm changes, and other noise.

That is what makes it so useful now.

When the whole market is shifting, you need a way to ask, "What happened because of our change?" Not "what happened while our change was live?"

Those are very different questions.

At SearchPilot, this is exactly the kind of problem we care about. We want SEO teams to be able to tie a specific change to a specific impact, instead of trying to reverse engineer meaning from a messy chart after the fact.

In the webinar, I talked about a real customer test where we increased organic clicks into PDPs by 17.5%. Think about what that means if those PDP clicks are from users who are already close to buying.

That is the kind of result that can change how leadership sees SEO.

Multi-metric testing matters now

One of the reasons I am excited about this area is that SEO measurement can no longer stop at "daily organic sessions from Google."

That number is too blunt.

With multi-metric testing, we can look at different types of search visibility and behavior. For ecommerce sites, that might mean Merchant Center clicks, Merchant Center impressions, PDP clicks, revenue, and other commercial signals.

That matters because the same change can have different effects across different parts of the search experience.

A change might not move classic organic clicks much, but it might improve product grid visibility. Or it might increase clicks to PDPs while reducing low intent traffic elsewhere. Or it might improve traffic but hurt conversion, which is not a win.

This is why SEO teams need to think beyond one metric.

Search is no longer one surface. Measurement cannot behave as if it is.

Full funnel testing should be part of the cycle

SEO changes can affect user experience.

That sounds obvious, but SEO teams and CRO teams have not always tested together. Sometimes SEO teams make changes to improve visibility. Sometimes product or CRO teams make changes to improve conversion. Each side worries about the other side later.

That is risky.

If you add content to a page for SEO, users still see it. If you remove content for conversion reasons, search engines still see that it is gone. If you move modules, change copy, adjust templates, or change PDP structure, you may affect both acquisition and conversion.

So test both.

At SearchPilot, we call this full funnel testing: SEO testing to measure the search impact, and CRO testing to understand what happens when users land on the page.

That is especially important now because the remaining click is more valuable. If you have fewer but more motivated visitors, you cannot afford to make the landing experience worse.

LLM referral testing is coming into view

One area to start preparing for is LLM and AI referral testing.

Commercial purchasing directly inside AI assistants is not fully here yet. Universal commercial protocol is not something teams can rely on today. People still need to click somewhere when they want to make a purchase.

That gives ecommerce brands a window.

Start asking whether site changes affect AI referrals. Start looking at how your products are being described. Start thinking about whether page content, product data, and brand information are helping AI systems understand and recommend what you sell.

The data is still early. The tools are still changing. But this is the right time to build the habit of asking testable questions.

Do not wait until the entire dashboard changes again.

What I want SEO teams to take away

The old SEO dashboard is not useless. It is incomplete.

If you still judge ecommerce SEO mainly by year on year clicks, you may be underreporting the value of the work. Worse, you may push the team toward the wrong goals.

A low intent click that used to happen earlier in the journey may disappear. That hurts the traffic chart. But if a higher intent click still arrives at the PDP and converts, the business may be doing fine.

That is the story SEO teams need to get better at telling.

Clicks are not all equal. A PDP click from a shopper who already knows what they want is not the same as a casual research click to an evergreen article. A Merchant Center impression is not the same as a blue link impression. A product grid interaction is not the same as an old SERP visit.

If shopper behavior is changing, KPIs need to change with it.

Do not chase the old signal if it no longer maps to the business result.

Put search in control mode with SearchPilot

There is a reason this topic matters so much to us at SearchPilot.

Search is often the biggest channel and the least understood. When the old reporting breaks, teams can either argue over what the numbers might mean, or they can test their way to a clearer answer.

SearchPilot makes SEO and GEO testable so teams can move from guessing to knowing. We run controlled experiments across category pages, product detail pages, navigation, content, internal linking, and other high value site sections, then show the impact with timelines and confidence.

For ecommerce teams, that increasingly means looking beyond total traffic. PDP clicks, Merchant Center clicks and impressions, conversion rates, revenue, and AI referrals all need to be part of the conversation.

The goal is not to predict every search behavior shift before it happens. Nobody can do that cleanly.The goal is to build a system that can learn as the market changes.

Stop trying to predict the future. Experiment to discover it. If you want tailored test ideas for your top PLPs and PDPs, schedule a demo and we will share a starter list and a clear path from validation to velocity to control.