What to do now that AI Overviews turned search into reading sessions

Intent still tells you what to write. But erstwhile an AI Overview (AIO) lands connected the SERP, users nary longer behave the aforesaid measurement arsenic classical search. 

In this memo:

• AIO compresses 5 chopped hunt intents into 1 reference pattern, and the 20-year SEO presumption it breaks.
• What winning the “second impression” looks for illustration for product, category, and blog pages.
• The one-slide mentation that reassures stakeholders the contented team’s past 3 years were not wasted (premium).
• A Claude accomplishment that audits your meta descriptions against the competitors sharing your SERP (premium).

View embedded content

The caller intelligence exemplary of hunt intent

Last week, I shared really Eric Van Buskirk of Clickstream Solutions and I analyzed anonymized clickstream information from astir 846,000 U.S.-based Google hunt sessions.

The astir important finding? The time-on-page for a personification connected the SERP is nary longer limited connected hunt intent erstwhile an AIO is present. The AIO compresses hunt intent behaviour to look akin crossed intent types. 

• Old intelligence exemplary of hunt intent: Navigational searches are “fast.” Informational ones are “slow.” Time-on-SERP follows intent, and SERPs without an AIO intelligibly show this shape (similar to classical hunt pre-AI outputs), demonstrated by 12% of navigational searches vs. 32% of section searchers still being connected the SERP aft 21 seconds.
• New intelligence exemplary of hunt intent: There is hardly a quality successful really agelong users walk clip connected SERPs betwixt personification intents erstwhile location is an AIO present. 42-48.5% of users are still connected the SERP aft 21 seconds crossed each the 5 awesome intents.

From “The aforesaid personification behaves otherwise successful AIOs vs. AI Mode” (bold added):

At 21 seconds into a convention without an AIO, only 12% of navigational searchers are still connected the page. 32% of section searchers are. In classical search, time-on-page has ever followed intent: navigational users time off accelerated because they cognize wherever they’re going, section users enactment because the SERP is dense pinch maps and listings, informational users autumn location successful between.

With an AIO present, the dispersed compresses to hardly 6 points. All 5 intent types(informational, local, navigational, transactional, video) cluster betwixt 41.9% and 48.5% time-on-page astatine 21 seconds.

Notice really overmuch longer mean SERP sessions are — almost 4x! So, we tin reason that AI Overviews don’t conscionable compress personification intent but besides prolong the clip users walk pinch hunt results. 

The reason? Additional context. Direct answers from the AIOs supply much accusation and return longer to read. The intent down the first query matters less.

This is the spread betwixt links and answers successful the caller AIO-filled SERP. In the past, giving users a database of (10 blue) links meant the personification was responsible for verifying accuracy and uncovering the accusation aft the click through. Therefore, Google gets personification feedback from their behavior. 

But erstwhile Google (or different LLMs) gives the reply directly, that onus is connected the reply engine.

Bing’s blog, “Evolving the domiciled of the index,” brings this to a point:

Grounding an AI–generated reply introduces a fundamentally different constraint: The strategy is nary longer conscionable pointing to information, it is utilizing it. The extremity shifts from “fetch the champion documents” to “fetch the champion accusation to synthesize into a reliable, verifiable answer.”

Lastly, it besides intends location is inferior successful search branded prompts much diligently and making judge LLMs return the desired accusation astir a brand. Just for illustration companies bid connected their marque arsenic a defense mechanism, they should show branded prompts, not conscionable product- aliases painpoint-related ones.

Heads up: The Adapting for AI-Mode Based Search deck gives board a validated, data-backed communicative to coming to executives who support asking what changed. Find it successful the Premium Subscribers Resource Library.

Why this matters

For 20 years, what you searched told Google and SEOs really you’d behave. Type a marque sanction (navigational search), and you’re successful and retired successful seconds. Search “best CRM for startups” (comparison search), and you settee into a group of comparison pages. Intent sorted everyone.

The AIO erased that tell. By dropping a artifact of reply matter astatine the apical of the page, it pulls each searcher into a reference session, nary matter why they came. The brand-name searcher sounds the AIO. The merchandise interrogator sounds another. Both slow down, some stay, some behave alike connected the SERP page. That flattening is the intent compression.

Most Google users ne'er chose this, because astir Google users are not AI early adopters. They meet AI done Google’s hunt results arsenic Google forces guides them into AIOs and AI Mode astatine the apical of the results. AI is changing its hunt behaviour passively and without searchers’ definitive consent. Many Google users mightiness not moreover recognize they’re utilizing AI, and that’s 1 logic we’re seeing growing installs of DuckDuckGo.

Google reports more than 1.5 cardinal people usage AI Overviews, truthful this is not an separator case. It’s really the web is searching now.

What winning the 2nd belief looks like

The AI Overviews vs. AI Mode behaviour analysis revealed the value and optimization opportunity of the 2nd look. The 2nd belief is what searchers spot connected the back-scroll, aft they’ve already passed your listing once. It’s for illustration a double-take made by a market shopper successful the cereal aisle who has dozens of options: The shopper scans each container successful view, past circles backmost to reread the 1 that caught their eye.

Metadata is the trigger for selecting hunt results. Rich snippets drawback attraction early on, but they mightiness not beryllium capable to person users to a click, particularly if caller hunt behaviour has shifted to see a thorough reference of the SERP and a 2nd scroll up. What tin gain the click is what shows up adjacent to your listing connected that 2nd pass, and it has to beryllium applicable and trustworthy.

Different page types require different relevance and spot indicators.

Product item pages (PDP)

Users comparison prima ratings, reappraisal count, price, and banal status.

Three things to power extracurricular the meta:

• Product schema pinch aggregateRating, review, offers, and availability. Miss immoderate 1 of these, and a competitor’s listing renders fuller than yours.
• Review count is simply a comparison field. 47 reviews adjacent to a competitor’s 2,300 loses connected the 2nd walk moreover if your explanation is sharper. Review velocity is simply a competitory moat.
• Multiple images successful the schema array truthful Google has options for different SERP layouts.

Category item pages (CDP)

Category pages compete pinch the AIO’s ain list. If the AIO already enumerated 5 options, your class page has to look for illustration the spot wherever the personification goes to really take betwixt them.

Three things to power extracurricular the meta:

• ItemList schema connected the class page truthful Google tin render merchandise carousels successful the SERP. A carousel takes much vertical abstraction than a azygous listing and dominates the back-scroll.
• Filter and benignant UI visible successful the SERP preview. Google sometimes surfaces sitelinks for class facets (“by price” aliases “by brand”). Internal linking to those facets makes them eligible.
• Page count and depth. A class page pinch 12 products competes severely against 1 pinch 240. The 2nd belief carries an implicit “Is this comprehensive?” check.

Blog content

The AIO has already fixed the personification the answer. To gain a imaginable validation click aliases astatine slightest a thorough 2nd belief for your brand, what the personification is looking for is credibility connected who said it and when.

Two things to power extracurricular the meta:

• Visible datePublished aliases dateModified successful the SERP. A 2024 article adjacent to a 2026 article apt loses, whether the personification considers the explanation aliases not.
• Article schema pinch a named writer field that links to a sameAs URL (LinkedIn, writer bio page). This makes the writer an entity Google tin resolve, which matters for E-E-A-T scoring moreover if nary visible paper renders.

The AI SEO Change Management Plan alone saves board 10+ readying hours for retraining your squad against existent champion practices, but Premium is only $150/year. See the afloat library.

What intent compression intends for operators

The past 3 years of intent-based contented activity produced the correct pages for the correct queries. What’s shifting is 1 prediction furniture connected apical of that strategy: really agelong users stay, wherever they look, erstwhile they click.

That furniture is now AIO-driven, not intent-driven. User hunt intent still drives what your marque needs to write, but it is nary longer a bully estimate of how users will behave connected a SERP page.

Therefore, much optimization activity moves to the SERP, focusing really your listing sounds against the AIO and the results astir it, abstracted from the intent logic that decides what the page contented should beryllium to reply the query. 

Your halfway optimization efforts don’t change. A contented squad built astir intent clusters keeps its cluster representation intact. What gets updated is the optimization shape per page (meta descriptions, title tags, the second-impression framing from finding 7 shared past week pinch Premium subscribers), not the underlying taxonomy aliases contented strategy.

This station first appeared connected the author’s website and is republished present pinch permission.