Savannah Young Research

Remitly · Discovery and search behavior

Where the category gets decided now

Marketing needed to know what people actually type into AI tools when they are choosing how to send money abroad, and what that means for the content strategy.

Method
Unmoderated qualitative
Sample
8 participants, North America, UK, Australia
Corridors
India, China, Brazil, Somalia, and others
Collaborator
Adam Whittles

What I designed

Unmoderated sessions with current customers and non-customers who send money abroad, personally or for business. Participants walked through their recent real use of generative AI tools to research transfer options, fees, timelines, and safety, in their own words and their own prompts.

Directional by design. The study was run to capture prompt language, not incidence.

A limitation I wrote into the report rather than leaving to the reader: the sample was experienced senders, so the prompts skew toward rate, speed, and safety. The beginner question, how do I even do this, is not represented here.

What I found

Search engines had been demoted to a verification step. Discovery, comparison, and shortlisting were happening inside AI tools, and people went to Google afterward to check the answer rather than to find one.

The trust behavior was the most useful part. People asked the tools to summarize customer reviews, and some asked specifically for the worst ones, on the theory that weaknesses are where you learn something. Answers were trusted more when they carried citations, visible dates, and links out.

It gives me reviews, fees, options, all in one go. Then I go to the websites to confirm.Participant, Canada to India

Competitors had already claimed the adjectives the tools repeat back. One was described consistently as having the lowest fees and transparent pricing, whether or not that was true on any given corridor.

What it changed

Content strategy moved away from keywords and toward question-and-answer structure built for AI retrieval: country-pair comparison pages, published answers with schema markup, visible dates and author credibility, and an owned approach to review summaries.

It also raised a positioning question nobody had been asking. These tools are already describing you to people who have never heard of you. Which two attributes do you want them to use?

← All work Measuring $20M of brand spend →