One year of AI search: Insights for travel, food, & personal care

Odds are, your organic traffic hummed along, steady enough to lull everyone into a familiar SEO daydream: We’re doing the right things, the dashboard agrees, the universe is stable. 

Then May hit, and things got weird.

Impressions held, and the demand was still there. But clicks started thinning out in a way that was hard to explain, and harder to ignore.

By September, the pattern wasn’t subtle anymore. Impressions dropped, clicks dropped, and suddenly the old comfort blanket of “visibility = visits” stopped working.

Right now, search is starting to behave less like a list of options and more like a single, confident response.

That shift doesn’t mean websites stop mattering, but it does mean they get used differently.

A lot of the time, your content still powers the answer… but it doesn’t always earn the click. So the win condition expands: not just “Did we rank?” but “Did we get chosen, cited, and trusted?”

Over the past several months, the DEPT® SEO team has been tracking where attention actually goes when search includes LLMs, and what those systems decide to surface, cite, or quietly absorb without sending anyone to a website. 

That means we’re doing less rank watching and more investigative work: spotting which pages are getting picked up but rarely referenced, learning what language and formats get pulled into answers, and finding better ways to show stakeholders when the brand is being reached through AI, even if the session never happens.

We focused on three industries upfront—travel, food, and personal care—because they sit within the same search ecosystem but come with very different expectations. And those expectations are where the click patterns start to make sense.

A quick note on the dataset: This analysis reflects what we saw across six domains we monitored in travel, food, and personal care. It’s a small sample by design, so treat the takeaways as signals and patterns, not universal benchmarks.

Unlike travel (where AI referrals are at least inching up), food sits at a low share of AI traffic, and it’s trending downward. 


Food queries skew instant gratification, so if the answer is a quick substitution, cooking time, or “What can I make with chickpeas,” AI can satisfy it on-platform. And if the user never clicks… traffic goes down. 

Within that shrinking pool, ChatGPT’s share rockets upward, while other AI platforms are flat or falling, meaning even when total AI traffic is softening, ChatGPT is increasingly the gatekeeper for what food-related content gets a chance to earn a click-out.


When AI does send traffic for food, it’s increasingly favoring informational content over recipes. 


What that says about behavior:

• Users still ask recipe-style questions, but they don’t always need to visit a recipe page to get the gist.
• Informational pages (guides, tips, explainers) are more “quoteable” and easier for AI to reference as a source of truth.

AI isn’t killing recipes, but it’s re-weighting what earns the click. The how/why/what content is becoming the front door, while recipes fight harder for the visit. Your best chance to win the remaining click-outs is to own the informational layer AI prefers to send traffic to, and make recipe pages “AI-citable. 

Personal care AI-referral traffic is a relatively small share overall, and it shows a seasonal dip rather than a clean growth story. AI isn’t pouring traffic into personal care sites consistently, and it might be tied to seasonal intent shifts and category demand spikes. 


Like travel and food, ChatGPT is also the steady winner for personal care traffic.


But unlike travel and food, AI traffic skews toward product detail pages. Product detail pages account for the largest share of AI traffic across multiple AI sources, with over 40%. Meaning, in personal care, AI is actively nudging users toward specific products. Informational/advice also holds a decent share, about 27%. 


Personal care doesn’t behave like food (quick answers, fewer clicks) or travel (big planning journeys). It’s a two-step dance: trust-building + product decision, often in the same breath.

In our data sets, we found that conversions are slightly lower than traditional organic search conversions. 

Travel: 
Organic conversion rate: 1.20%
AI conversation rate:  0.94%

Food: 
Organic conversion rate: 32.83%
AI conversation rate: 32.18%

Personal care: 
Organic conversion rate: 76.43%
AI conversion rate: 72.81%

The gaps aren’t catastrophic, but they are a signal. AI is still heavily serving the discovery layer, and users often arrive earlier in their decision journey than they do from classic organic. The opportunity now is to close that last-mile distance with clearer next steps, stronger trust cues, and content that bridges intent to action. 

A content format that’s starting to matter: “answer packets.”

If AI is pulling from you, make it easy to pull from you. Build small, verifiable modules for high-value topics, e.g., FAQs, policies, product claims, safety guidance, comparisons, with clear headings, consistent phrasing, and sources/links. The goal is to be the most quotable, reliable version of the truth on the topics customers keep asking about.

The old objective was to earn visibility, clicks, and growth. The new story is messier, but honestly, more interesting. AI keeps a lot of the early-stage attention on-platform, and only hands off a click when it needs something the assistant can’t deliver alone like proof, depth, comparison, purchase details, and reassurance.

And when AI does send traffic, it tells you exactly what kind of value your category needs to provide. In travel, people still click because the stakes are high and they want receipts. In food, the click is often optional because the answer is enough. In personal care, AI behaves like a product concierge, splitting attention between reassurance content and product detail pages where decisions happen.

The conversion story backs it up: AI sessions can be slightly lower converting because they often arrive earlier. This is an opportunity to shorten the distance between “help me understand” and “help me choose,” with clearer next steps, stronger trust cues, and content built to support decisions.

Which is also why measurement has to evolve. Rankings and organic sessions still matter, but they don’t tell the full story anymore. The new scoreboard includes things like: share of citation, accuracy of how you’re represented, and handoff conversion (when someone moves from an AI answer to a meaningful action on your site).