AI Overviews collapsed organic discovery, and the traffic didn’t disappear — it routed through aggregators that resell your own customers back to you at a 25% margin markup. Ahrefs measured a 58% drop in top-ranking click-through rates, and Pew Research found 26% of users ending their session entirely after the AI summary. The displaced demand is being captured by comparison sites and affiliate networks. Without attribution that distinguishes intercepted customers from genuinely new ones, you’re paying twice for the same buyer.

The Discovery Layer Just Moved

The structural shift happened faster than most stores’ analytics caught up to it. AI Overviews — Google’s generative search summaries that appear above traditional results — measurably reduced clicks on the underlying pages. Ahrefs’ 58% CTR drop on position-one listings is the headline, but the more interesting figure is from Pew Research Center: 26% of users ended their session entirely after seeing an AI summary, versus 16% without one.

That session-ending behaviour is what moved the discovery layer. Users who would have clicked a result and started a comparison journey now read the summary and make their decision — or abandon it — without ever leaving Google. The traditional organic discovery funnel for WooCommerce stores got compressed at the top.

The displaced demand had to go somewhere. It went to aggregators. Comparison sites, “best of” content publishers, affiliate networks, lead-generation platforms, deal sites, niche review properties — all of them positioned themselves to capture the search traffic that no longer flows through traditional organic channels.

The customer who used to find your store directly through search now finds an aggregator that finds your store on their behalf.

The 25% Margin Markup

The economics of the aggregator layer are simple, and brutal for stores that aren’t paying attention. Most affiliate networks operate on commission structures in the 15–35% range, with 25% as a common median for ecommerce categories. Comparison sites that monetise through affiliate links use the same fee scale.

That 25% comes off your margin every time the aggregator ranks for a query, the user clicks through, and the conversion happens. The aggregator does no inventory work, no fulfilment, no customer service, no product development. They’re a discovery layer that exists because the original discovery layer — direct organic search — got compressed by AI Overviews.

The crucial question is whether the customer the aggregator delivers is genuinely incremental. Did the aggregator introduce a buyer who would never have found you otherwise? Or did the aggregator intercept a customer who was already in your funnel — searching for your brand, comparing against you, or returning to complete a purchase — and slot themselves into the conversion path to collect a commission?

Stores without first-touch attribution treat both cases identically. The aggregator gets paid 25% either way. The first case is a real acquisition cost. The second case is margin leakage on a customer the store already had.

The Three Categories of Aggregator

The aggregator layer isn’t one thing. It’s three structurally different categories, each capturing search demand differently and each with its own incrementality profile.

Comparison sites. Properties that publish “best [category] of 2026” content optimised for the queries AI Overviews compress. These tend to capture genuine top-of-funnel research traffic, with mixed incrementality depending on whether the user knew the store before reading the comparison.

Affiliate networks. CJ, Awin, Impact, Rakuten, ShareASale, and similar platforms manage publisher relationships at scale. The incrementality varies wildly across the publisher base — coupon sites, in particular, often intercept rather than introduce — and the network’s reporting tools rarely distinguish the two.

Content networks and deal aggregators. Newsletter promotions, deal sites like Slickdeals, niche review blogs that double as affiliate publishers. The newsletter and deal model often introduces genuinely new customers; the niche review model is harder to assess without underlying touch-path data.

You may be interested in: Click-to-WhatsApp Ads Are Your Biggest Attribution Black Hole

The Signal Required to Detect Cannibalisation

Detecting which aggregator-driven sales were intercepted versus introduced needs a specific data architecture. Three things have to be true:

1. Every visit to the store is captured, not just visits that end in conversion. A customer who visits directly today and purchases via affiliate two weeks later needs both touches recorded.
2. The visit history is associated with a stable user identifier that survives across sessions, devices, and channels. Browser cookies alone won’t do this — they get cleared, expire, and fragment.
3. The affiliate conversion is matched against the user’s prior visit history at the moment of attribution, not just at the moment of click.

That combination of requirements rules out most stock affiliate tracking. Affiliate platform pixels see the affiliate click and the conversion. They don’t see the user’s prior direct visits, branded searches, or organic page views. The cannibalisation is invisible because the data needed to detect it sits outside the platform’s view.

It’s not invisible to the store, though — if the store collects the data first-party. A server-side pipeline that captures every page view, associates it with a stable identifier, and writes the touch sequence to the store’s own warehouse has all the information needed to reconstruct the pre-affiliate journey.

The aggregator can’t see your full funnel. You can. Whether you do depends on the architecture.

The Incrementality Question, Operationalised

Once the data exists, the analysis is direct. For each affiliate conversion in a given period, segment by first-touch:

• First-touch is the affiliate and there are no prior visits — genuinely new customer, full commission earned.
• First-touch is direct, branded search, or organic — customer was already in your funnel; the affiliate intercepted.
• First-touch is paid (Google/Meta/etc.) — customer was acquired by paid spend; the affiliate is double-commission territory.
• First-touch is a different affiliate — networks are competing to last-click each other; the introduction value sits with whichever publisher actually originated the customer.

The percentage in each bucket gives the store its actual incrementality figure, and the actual incremental cost-per-acquisition. Stores that run this analysis usually find the intercepted bucket is 30–60% of total affiliate conversions, depending on category and program design. That’s the margin leakage figure that should be informing program negotiation, publisher allowlisting, and commission structures.

You may be interested in: The Eight Hops a WooCommerce Conversion Has to Survive

Here’s How You Actually Build This on WordPress

Transmute Engine™ is a first-party Node.js server that runs on your subdomain, capturing every WooCommerce page view and event into a stable customer record before any third-party platform sees the data. The full touch sequence — direct visits, branded searches, paid clicks, affiliate clicks, and the eventual conversion — lands in your own BigQuery dataset for incrementality analysis. The aggregator economics stop being invisible because the architecture stops sending fragmented touch data to platforms whose interest is in counting their own clicks.

Key Takeaways

• AI Overviews compressed traditional organic discovery — Ahrefs measured 58% CTR drop on top results, and Pew found 26% of users ending sessions on the SERP.
• Aggregators captured the displaced demand and now resell it to stores at typical 25% commission structures.
• Most affiliate conversions are a mix of incremental and intercepted; without first-touch attribution, stores can’t tell which is which.
• Affiliate platform pixels can’t detect cannibalisation — they see the affiliate click and conversion but not the user’s prior visit history.
• First-party server-side tracking is the architecture that captures the full touch path needed to compute real incrementality.

Frequently Asked Questions

If you can’t see which affiliates introduced new customers versus intercepted existing ones, you’re paying for the same customer twice. Seresa builds the architecture that makes the difference visible.