Designing for Distributed Trust:
Metasearch and Flight-Booking

* Specific details have been omitted or modified for proprietary reasons.

ROLE:

PARTNERS:

SCOPE:

Research Lead

Shopping,
MileagePlus, Loyalty,
CX leadership

Diary study with 47 recent flight bookers using real bookings with screen recording and live narration

Presented to VP and senior leadership across Shopping and Loyalty as part of a broader C-suite priority on direct booking

Informed direct booking strategy and meta-handoff design.

Problem Space

Digital purchase journeys increasingly begin outside the company's owned experience. Customers compare options through search engines, aggregators, marketplaces, credit-card portals, social recommendations, and now AI tools that summarize or interpret choices on their behalf. The company may still own the transaction, but independent channels increasingly shape the customer's first sense of value, fairness, and confidence.

Flight shopping makes this problem especially visible. Customers rarely evaluate a fare in isolation. They compare departure times, arrival windows, airports, layovers, baggage rules, fare restrictions, loyalty benefits, and the possibility that a better price may appear later. Many of those judgments now happen in metasearch environments — Google Flights, Kayak, Skyscanner, Expedia — before customers ever reach an airline site or app.

United had observed a meaningful pattern: customers entering from metasearch were converting at higher rates through downstream booking steps than customers who began directly on United's channels. Meta-originating customers often bypass Flight Search Results and enter later in the booking flow, but the behavioral significance of that bypass was uncertain.

The core question was whether metasearch customers were arriving at United having already done decision work that direct customers still had to do inside the airline flow. If confidence was being built outside the owned experience, the airline still had to understand what kind of trust was being created there and what kind still needed to be earned at purchase. As AI becomes more active in planning and comparison, that distinction matters more, not less: companies need to know which parts of trust they can own, which parts they have to mirror, and which parts they have to verify.

Summary

Flight shopping increasingly happens across a distributed set of channels — metasearch platforms, airline sites, loyalty programs, credit-card portals, and emerging AI tools that interpret options before customers ever reach the airline's owned experience. Internal data showed that customers entering from metasearch were converting at higher rates through downstream booking steps than customers who began directly on United's channels. The pattern was clear; its cause wasn't. I led a diary study with 47 recent flight bookers to understand how customers actually compare flights, decide when to stop searching, and judge whether an itinerary is good enough to buy. The research showed that comparison channels and airline channels support different kinds of trust: metasearch helps customers establish market confidence by surveying the field, while airline sites help them establish purchase confidence by confirming what's actually being bought. The conversion ga wasn't a channel-preference problem. It was a decision-readiness problem — and one that points to where direct booking strategy needs to go as planning increasingly happens outside the airline's owned experience.

Approach

The research used a diary-study format in which participants booked real flights they actually needed. They completed the study on either mobile or desktop, depending on how they normally shop. During the session, they recorded their screen and narrated their reasoning aloud as they searched, compared, narrowed, selected, and booked.

The study was structured around three moments in the decision: a starting state (what they had already researched and decided, what still needed to be confirmed), live booking (which tools they used, what created hesitation, what made an itinerary feel good enough), and immediate reflection (what ultimately allowed them to book, what remained unresolved, where ancillary decisions fit).

Participants used their normal tools rather than a prescribed platform. These included Google Flights, Kayak, Skyscanner, Expedia, airline sites and apps, credit-card travel portals, and occasional AI tools. Quantitative measures captured tool usage, ranked decision factors, sources of difficulty, confidence drivers, and source of reassurance. Open-ended responses explained how those signals appeared in practice.

The design choice that mattered most was letting participants use their own real bookings rather than constructing a controlled task. Flight shopping is shaped by tradeoffs — schedule, price, airport, family logistics, loyalty status, work calendar — that don't surface cleanly in a hypothetical task. Anchoring the study in real trips meant the friction, the hesitations, and the stopping rules came from actual decisions with real stakes. That's what made it possible to see why customers were doing what they were doing, not just what tools they used.

What We Found

Customers Were Trying to Decide When to Stop Comparing

Across the study, price and timing dominated the decision, but rarely as a hunt for the absolute lowest fare. Customers evaluated whether a cheaper flight created another cost: a worse airport, an early departure, a long layover, a missed work window, a rental car, a riskier family itinerary. Flight choice was a tradeoff problem before it was a checkout problem.

The most common difficulty wasn't lack of options. It was balancing competing priorities — and a meaningful share of participants specifically reported difficulty figuring out which option was good enough to stop comparing. Customers weren't stuck because they had too few options. They were stuck because too many options remained plausible.

The clearest evidence of this came when stopping rules broke down. One participant couldn't complete the booking because prices had risen sharply from a fare she'd seen earlier, making every current option feel like a mistake. Another moved across Skyscanner, Copilot, Priceline, Air Canada, promo-code searches, and baggage rules because the trip had to satisfy price, work schedule, direct-flight preference, and bag constraints simultaneously. In these cases, more information didn't create confidence. The missing piece was a stable basis for commitment.

Metasearch Created Market Confidence by Making Alternatives Legible

Comparison tools played the largest role in narrowing. Most participants used Google Flights, Kayak, Skyscanner, Expedia, or another comparison tool more than any other source during the session. These tools helped customers see the market at once: airlines, airports, dates, departure times, layovers, price bands, and fare movement.

The strongest confidence signal was tied to this comparison work. The largest single share of participants said that seeing one option clearly stand out in a comparison tool helped them feel confident moving forward. The standout option wasn't always the cheapest. It could be the flight that best balanced price and schedule, avoided a connection, used a preferred airport, preserved a workday, or made a family itinerary easier.

What Changed

The research reframed the metasearch conversion gap as a question of decision readiness rather than channel preference. Meta-originating customers often reach the airline after comparison work has already been performed elsewhere. They've seen the market, evaluated alternatives, and identified the itinerary worth pursuing. Direct customers may still be doing that work inside the airline's own flow. Treating both groups as if they arrived in the same decision state had been obscuring the real opportunity.

That reframe is now reflected in how the Shopping team thinks about direct vs. meta-originating customers. The "market confidence vs. purchase confidence" framing surfaced a clearer principle: purchase confirmation is where the airline's authority is strongest, and that's where the company should be building confidence-creating capability that metasearch platforms structurally can't replicate. The work has shaped current thinking on what to mirror from metasearch (price context, fare inclusion clarity, schedule tradeoff legibility) and what to assert as authoritative (final total price, fare rules, loyalty value, transaction reliability).

Metasearch helped create what's best described as market confidence: the feeling that the customer had seen enough of the relevant market to know which option deserved attention. This is the form of confidence that direct experiences need to produce but often don't fully support when customers begin inside the airline's own search results.

Airline Sites Created Purchase Confidence by Confirming What Was Actually Being Bought

Comparison tools helped customers identify likely choices, but airline sites and apps carried a different kind of authority. Customers went to the airline to confirm final price, fare tier, seat availability, baggage rules, refundability, loyalty numbers, status benefits, payment, and confirmation.

Roughly the same proportion of participants who cited a standout comparison option as a confidence driver also cited confirming final total price and included benefits on the airline site. A separate large share cited understanding rules and add-ons on the airline site. Comparison confidence and purchase confidence weren't interchangeable. They were complementary.

Participants repeatedly described the airline site as the place where the decision became real. Some compared prices on Google Flights or Kayak but checked the airline site to make sure the same fare existed. Others used the airline site because they distrusted third-party booking, or because of past issues when a reservation or add-on didn't translate cleanly. Several described the airline site as the only source they would trust for rules, restrictions, bags, seats, or changes.

This confirmation role mattered most when benefits changed the value of a fare. Customers with loyalty status, credit-card benefits, or stored profiles often needed the airline site to confirm that the effective value was better than the comparison result alone suggested. A fare that looked similar in metasearch could become preferable once free bags, points, seat access, refundability, or account convenience entered the calculation.

But the airline site could also reopen the decision if important information appeared late. Seat fees, carry-on restrictions, separate-ticket warnings, or unexpected fare limitations made some customers pause after they'd already narrowed the choice. When that happened, the booking path shifted from confirming a decision to forcing the customer to reevaluate it.

Trust Was Distributed; Accountability Stayed With the Airline

No single source owned trust across the whole decision. When asked which source most reassured them that booking was the right decision, airline sites/apps and comparison tools were essentially tied. Past experience with the airline came next. AI was named by a small minority.

Customers trusted comparison tools for visibility across the market. They trusted airline sites for transaction accuracy. They trusted their own past experience for operational reliability. Each source had a different role.

The clearest evidence of this accountability appeared around add-ons and third-party booking. Most participants didn't purchase ancillary options. When add-ons did matter, they were functional rather than discretionary: bags, seat assignment for family travel, refundability, fare tiers that prevented later uncertainty. Comfort-oriented extras rarely shaped the original flight decision.

Participants overwhelmingly expected to handle these purchases through the airline. The reason wasn't convenience alone. It was confidence that the purchase would attach correctly to the booking record and could be resolved with the airline if something went wrong. Search platforms were useful for comparison; customers were reluctant to rely on them for anything that had to be honored at the airport.

AI played a smaller but strategically important role. No participant used AI as their primary tool, and only a small share said an AI summary or explanation helped them feel confident. But the cases where AI did appear were revealing — clarifying refundability, checking whether a better deal might exist, understanding a rule faster than searching policy pages. AI wasn't yet replacing metasearch or airline booking sites, but it was beginning to occupy the interpretive space between comparison and purchase.

The findings feed into a broader strategic priority on direct booking — one of the company's most consequential digital strategy questions. The research is being used to inform how the direct path can support comparison closure earlier (price context, schedule tradeoff clarity, fare inclusion transparency, cues for when customers have seen enough), how meta handoffs should preserve continuity rather than make customers restart their reasoning, and how AI-mediated shopping should be understood as a new layer between comparison and purchase rather than a replacement for either.

The work was presented to VP and senior leadership across Shopping and MileagePlus, and is now part of how the team thinks about a longer-term question: how to keep customers who increasingly form confidence through independent channels arriving at — and converting through — the airline's owned experience.

The broader takeaway, which travels beyond airlines: customer trust no longer lives in one place. Market trust forms where customers compare. Transaction trust forms where they buy. Relationship trust forms through prior experience with the brand. Strong purchase experiences will need to connect those forms of trust rather than assuming the owned channel can replace the rest of the decision ecosystem.