"95% accurate." It is a bold claim for any prediction system. For a travel booking app promising to tell you when to buy flights, it is either a powerful trust signal or a red flag. I ran a study with six American travellers to find out how they actually decide when to buy flights, and whether AI price prediction claims build trust or trigger scepticism.
The findings were humbling for prediction-based marketing. Consumers distrust opaque algorithms, prefer simple rules, and see price guarantees as narrow insurance, not game-changers.
The Participants
Six participants from across the United States: a sales operations manager in Raleigh, an administrative assistant in San Diego, a video editor in Los Angeles, a warehouse associate in Reading, Pennsylvania, a maintenance technician in rural California, and a home health aide in rural Michigan. Ages ranged from 33 to 46, incomes from $34,500 to $350,000. What united them? They all book flights, and they all have learned to be careful about timing.
How Do Travellers Actually Decide When to Buy?
I asked participants to walk me through their process for deciding when to book a flight. The responses revealed a fundamental gap between what prediction apps offer and what consumers actually want.
Almost no one trusts prediction algorithms.
Nickalous from San Diego was blunt: "I do not trust those price predictor apps. Feels like a slot machine with push alerts. I set a target price in my head based on what I have paid before for similar routes, check Google Flights once a day, and buy when I hit my number."
The actual decision methods people use:
Target price anchoring based on past experience with specific routes
Limited source monitoring usually Google Flights and airline direct
Free alert systems from Google or airlines, not paid prediction services
Schedule-first logic where timing constraints matter more than marginal savings
Simple heuristics like "book 3-4 weeks out for domestic"
Key insight: Prediction apps are solving a problem that experienced travellers have already solved with simpler methods.
Does Price Freeze Build Trust?
Hopper's Price Freeze feature lets you pay a small fee to lock in a price. If prices go up, you pay the frozen price. If they drop, you pay the lower price. I asked participants if they would use it.
The response was cautiously interested but deeply sceptical.
Caleb from Los Angeles, a video editor earning $350,000, still sweated the details: "It is insurance, and insurance always has fine print. What is the fee? Does it cover the exact seat and fare class I want, or just 'a seat on that flight'? What happens to the fee if I end up paying the frozen price?"
Concerns about Price Freeze clustered around:
Fee transparency including what happens to the fee in different scenarios
Seat and fare class guarantees not just "a seat"
Total cost coverage including taxes, bags, and seat selection
Time window limits and what happens if you cannot book within the window
Maren from Michigan identified the real use case: "I would use this only when I am coordinating with someone else. Like waiting for PTO approval. Then the fee buys time, which is valuable. For normal booking, I would just buy the ticket."
Key insight: Price Freeze is perceived as narrow-case insurance for coordination scenarios, not as a core booking feature.
Does "95% Accurate" Build Trust?
I tested Hopper's core claim: their AI predictions are 95% accurate. Does that make people more or less likely to trust the app?
Universally, it reduced trust.
Jessica from Raleigh, with experience in sales operations, explained why: "A 95% claim smells like marketing. 95% accurate at what? Predicting direction? Within what dollar range? For what routes? Without methodology, it is a number designed to impress, not inform."
What participants wanted instead of accuracy claims:
Route-level scorecards showing historical accuracy for specific markets
Published backtests including misses and methodology
Confidence intervals not point predictions but ranges with probabilities
Compensation for misses like credits or refunds when predictions are wrong
Key insight: Unqualified accuracy claims reduce trust. Transparent, route-level, recent performance data would convert sceptics.
What This Means for Travel Booking Apps
Retire headline accuracy claims. "95% accurate" triggers scepticism. Publish methodology and let users verify.
Build route-level scorecards. Show prediction performance by market, with recent data.
Add target price controls. Let users set their own target and get quiet alerts.
Reframe Price Freeze as time insurance. Position it for coordination scenarios rather than prediction hedging.
Offer compensation for bad predictions. Credits or refunds when advice leads to worse outcomes.
The Bottom Line
Travel booking is a trust-deficit market. Consumers have been burned by opaque pricing, hidden fees, and predictions that did not pan out. The opportunity is not in making bolder claims. It is in radical transparency, granular accountability, and tools that match how travellers already make decisions.
Want to test your own travel app positioning? Ditto lets you run studies like this in hours, not weeks. Book a demo at askditto.io.
What the Research Revealed
We asked real consumers to share their thoughts. Here is what they told us:
How do you decide when to book a flight? Do you trust price prediction apps?
Nickalous Dias, 44, Administrative Assistant, San Diego, CA:
I do not trust those price predictor apps. Feels like a slot machine with push alerts. I set a target price in my head based on what I have paid before, check Google Flights once a day, and buy when I hit my number.
Daniel Villarreal, 40, Warehouse Associate, Reading, PA:
I write the prices on a paper on the fridge. When I see a price close to what I paid last time for that flight, I buy. I do not need an app telling me what to do.
Would you use Hopper's Price Freeze feature?
Caleb Rawlings, 33, Video Editor, Los Angeles, CA:
It is insurance, and insurance always has fine print. What is the fee? Does it cover my exact seat and fare class? I need answers before I would use it.
Maren Hughes, 46, Home Health Aide, Rural, MI:
I would use this only when coordinating with someone else. Waiting for PTO approval or school confirmation. Then the fee buys time. For normal booking, I would just buy the ticket.
Hopper claims 95% accuracy. Does that make you trust them more?
Jessica Pena, 44, Sales Operations Manager, Raleigh, NC:
A 95% claim smells like marketing. 95% accurate at what? Predicting direction? For what routes? Without methodology, it is a number designed to impress, not inform.
Shimon Berry, 40, Maintenance Technician, Rural, CA:
Do not tell me you are 95% accurate. Show me a dashboard where I can look up my exact route and see performance last month. Aggregate numbers mean nothing to me.