Optimizing the Hotel Online User Experience

Introduction

I worked as a Senior UX Designer across IHG's brand sites — IHG, Holiday Inn, Holiday Inn Express and Crowne Plaza. My responsibilities covered visual and product design, competitor analysis, conceptual ideas, split-testing, analytics, and collaborating with the product team. The immediate team was product managers, a project manager, a data scientist and developers.

The work was continuous optimization rather than a single redesign: a standing testing programme where product hypotheses were researched, designed, built, run for two weeks and either rolled out or retired on the evidence.

Process and collaboration

Every test moved through the same six steps.

1. Product Owners provide customer painpoints, a test idea or a hypothesis explaining why they think it could work, their goals (what success looks like), and any constraints.
2. The core UX team determines whether the idea is relevant to the strategic goals of Web Channel leadership — revenue and bookings.
3. If yes, the UX team starts data or research insights. If it is A/B testable, we create the hypothesis and test briefs. This also entails business discovery, analysis of the test, and prioritization.
4. Design starts. UX tickets are submitted, a designer is assigned, concepts are created and approved, and test development begins.
5. Build starts. Coding and production development (target campaign, metrics validation and creation), dev testing, QA testing, and preparation for launch.
6. Run + analysis. Test monitoring for two weeks, results analysis, and a results deck created and shared with the Testing and Optimization team.

If successful, the test is implemented fully.

Tools and apps

Each phase of a test had its own instrumentation — qualitative session and interaction analysis going in, quantitative results analysis coming out.

The rate selection page repeated the pattern. Rate clarification was the largest category at 28%, followed by difficulty finding applicable special rates at 21%. The top three painpoints across the funnel were rate clarification, rates and room clarification — the same problem, described three ways.

Empathy map

Mapping what bookers were thinking, hearing, seeing and saying put the complaints in the user's own words. “Confused by the tax rates. Not clear what the total price per night is.” “The initial page has too many rate options and it is unclear what they mean.”

Personas

Two personas carried the research forward. Both list a version of the same painpoint — for John the price not matching the listing, for Tiffany confusion about room rates.

Competitive analysis

If rate transparency was the problem, it was worth knowing how the competition handled it. Marriott hides the full rate behind a Summary of Charges click and takes five steps to complete a booking. Hyatt displays the full rate details on load and takes four. Choice Hotels also takes four, but does not show full rate details on load — and pre-enrolls the user in its rewards programme, which registers an account. Only Hyatt was showing people what they were paying before being asked to pay it.

Design decisions and split testing

The payment page had been redesigned shortly before, to fix UX issues raised in user testing. The sign-in CTA moved into the form area and reservation details moved to the top of the page. But the layout had also become a single long column — which, for bookers who were not signed in, produced a page that required more scrolling than the design it replaced.

We ran three challengers against the control.

Outcomes

Experience C won. The streamlined page with rate details open by default consistently outperformed the control. On the IHG site, booking conversion increased 6% and revenue per visitor 5%. On the Holiday Inn Express site, booking conversion increased 5% and revenue per visitor 5%. Both were statistically significant.

The winning experience:

• Removed the top brand navigation bar and global navigation links
• Condensed the reservation summary
• Eliminated repetitive copy and excess spacing
• Increased the size of form section headers
• Flipped the payment section background to dark grey to differentiate it and visually emphasize security
• Moved the Geotrust and TRUSTe icons up from the footer into the payment section, to answer anxiety about card and page security where it occurs
• Reduced the size of the legal language
• Opened the rate details by default in the right column — on large viewports only, since on mobile's single column that would have undercut the effort to reduce page length

The device split was the more interesting finding. Desktop lifts came from a combination of content streamlining and rate details exposure, while mobile lifts came from streamlining alone. Bookers appear to want different things depending on the device: desktop users crave more information at checkout, mobile users want the basics and a quick exit.

The winner also drove stronger bookings with both audiences — anonymous visitor bookings up 5%, explicit member bookings up 3% — and is now served to all traffic on the IHG and Holiday Inn Express brand sites (US and UK English-speaking audiences) through Adobe Target.

The testing programme

Over 100 projects were tested in 2019 — 42% more than in 2018, and an average of eight a month. More than 75% of our A/B split-testing was successful across the IHG brands.

Learnings

I received a lot of pushback from stakeholders about the feedback coming from our users. Often it was met with disbelief. It was only when I collaborated with a data scientist and was able to summarize the findings and present them to the team that we got several people on board to make changes. It was a lesson I learned from other UX teams: data will always support design-driven decisions.

A/B split-testing turned out to be a powerful way to measure success — using data research, data analysis in Quantum Metric, competitor analysis and user testing as the foundation for deciding where to focus testing effort.

The other learning was about autonomy: maintaining the ability to test outside the normal web confines when needed, in order to improve what we learned and what we shipped.