Field notes · 30 April 2026

Travellers Don't Want Their 38 Tabs Replaced

Every AI travel planner I have looked at this year leads with the same pitch.

You know the one. Travel planning today means 38 browser tabs. Booking.com, Skyscanner, Reddit, TripAdvisor, GetYourGuide, three blog posts, two YouTube videos, a forum thread from 2019. AI will collapse all of that into a single chat window. Type your trip, get a plan, book it.

The data quoted alongside this pitch is real. Google and Expedia both find that the average traveler visits around 38 websites and spends more than five hours researching a single trip. Phocuswright's 2025 numbers show generative AI's share of trip research has roughly tripled in a year, from 6% to 15%, while search engines have dropped from 51% to 36%. There is a real shift happening.

The pitch still misdiagnoses the problem. The factual claims hold up. The diagnosis they sit on does not, and that is why most products built on it keep losing.

The Tabs Are Not the Problem

Tabs are a deliberate strategy.

The travel industry runs on commercial bias. TripAdvisor takes commission from the hotels it ranks. GetYourGuide ranks by margin. Blogs are sponsored. Instagram is performative. OTAs hide inventory their margin partners do not like. Every source the traveler touches has a financial reason to mislead them, and travelers know this. After a few burned trips they figure out the pattern.

So they triangulate. Three sources for a hotel, two sources for a price, a Reddit thread to sanity-check the blog. They are running a deliberate verification process across 38 tabs because they have learned, painfully, that no single source is trustworthy.

This is why every standalone AI planner runs into the same wall. It walks in promising to replace the tabs, and immediately inherits the trust deficit of a single source. A Booking.com survey found only 35% of users fully trust AI travel outputs. CNBC's reporting earlier this year cataloged the cases that earn that distrust: ChatGPT inventing a hotel in Tokyo, a Spanish influencer denied boarding because ChatGPT got the visa requirement wrong, AI tools sending tourists to landmarks that do not exist.

A Journal of Consumer Behaviour paper from earlier this year showed that even rare hallucinations measurably reduce user willingness to act on AI travel recommendations. Replacing the tabs is, in practice, a withdrawal of the verification system the user built to protect themselves.

Planning Is Also Pleasure

The second problem with the 38-tabs pitch is that it assumes planning is suffering people want to skip.

The research says something else. A 2010 study published in Applied Research in Quality of Life found that vacationers experience their largest happiness boost during the planning phase, not during or after the trip itself. A 2007 paper in the Journal of Experimental Psychology found people are happier anticipating a holiday than they are once they take one. The more actively a person plans, the more anticipation they feel, and the more lasting the satisfaction. Pre-packaged tours, where someone else has already done the planning, measurably reduce that anticipation.

There is a clean neurological explanation. Anticipation is a dopamine system. The reward fires while you are imagining the trip, not while you are eating the croissant. The hours spent across those 38 tabs are not all friction. A meaningful share of them are the highest-utility hours of the entire trip.

This is the IKEA effect, well documented in travel-planning research. Effort spent assembling something increases ownership of the result. The traveler who spent a Saturday building their own Japan itinerary feels it is their trip in a way that the traveler who clicked one button does not. Cut the assembly out and you cut out the ownership.

A planner that compresses the 38 tabs into one click is, for a meaningful share of users, deleting the most pleasurable part of the product they are paying for.

The Real Pain Is Bookkeeping

Decompose what is actually happening across those 38 tabs and the friction comes apart into five distinct, fixable pains.

Re-entering context. The dates, the budget, the travel companions, the dietary restriction, the no-red-eye rule. Each of these has to be typed again into every search box on every site. There is no shared spine for a trip.

Choice overload. Iyengar and Lepper's jam study showed customers presented with 24 jams bought 3% of the time, while customers presented with 6 bought 31% of the time. Travel sites stack thousands of options against a user with one decision to make. More than 50% of travelers abandon their booking flow once they reach it.

Comparing things that do not compare. A boutique hotel, an Airbnb, a hostel and a ryokan are not normalizable on any single axis, and no site does the work of normalizing them.

Trust calibration. Every source is biased differently, and the user has to recalibrate which source to believe for which kind of question. Reddit for honest food, blogs for hidden corners, OTAs for prices, embassy sites for visas. Holding that calibration in your head for five hours is exhausting.

No memory across trips. Preferences learned on the last trip do not carry to the next one. The same 38 tabs open again, the same restaurants get re-recommended, the same logistics get re-solved.

These five pains are what travelers are complaining about when they say "ugh, the tabs." None of them are about the existence of multiple sources. All of them are about the absence of a context spine that holds the work together.

What This Means for AI Travel Products

A planner that replaces the tabs fights human psychology on two fronts. It removes the user's verification system, so trust collapses. It removes the active assembly, so anticipation collapses. The result is what most current AI travel products turn into: pretty mood boards people open when they are bored, and close before they actually book anything.

A planner that orchestrates the tabs is a different product. It carries the context across sources without re-entering. It surfaces options from across the open web rather than substituting its own opinion for the user's research. It cites where every claim came from, so the user can verify on their own terms. It keeps the user in the role of planner-in-chief, with the AI doing the bookkeeping the user always hated. It remembers the user across trips.

The pitch under that product is not "no more 38 tabs." It is closer to: you still get to plan your trip, you still get the dopamine of building it, you still get to triangulate sources you trust. The AI carries the context, does the boring re-entry, and surfaces the trustworthy options across all the sources you would have opened anyway.

That is the version travelers do not push back on. It preserves the dopamine of planning and the verification habits that keep them safe. It only removes the bookkeeping.

What people complain about when they complain about the 38 tabs is the bookkeeping. Sell against the bookkeeping.