A single view of the member: CRM discovery for a luxury hospitality group

No tool bought on a pitch. No build started before the decision was made. The work produced one specification, a ranked backlog, and a scored vendor shortlist the client could defend to a board.

The client

A multi-site members-only hospitality operator, growing fast. The membership experience is the business: who a member is, what they hold, what they attend, what they buy. The systems carrying that knowledge were not keeping up with the growth.

The problem

The operation ran on more than ten disconnected tools. Membership sat in one system, bookings and events in a handful of others, point-of-sale in a third cluster, marketing and automation in yet more. None of them talked to each other cleanly. A member existed as several partial records in several databases, so no one had a single reliable view of anyone, and every new site multiplied the same fragmentation.

The client needed someone to establish, on evidence, what the real state was, what it would take to fix, and a recommendation they could put in front of a board and defend.

The work: discovery first, tools second

I started with an as-is audit. I inventoried the full tool stack and mapped it against the capabilities the business needed, producing a gap matrix that showed, tool by tool, exactly where the holes were. That single artifact turned a vague sense of "our systems are a mess" into a specific, six-category diagnosis anyone could read in a minute.

From there I ran structured interviews across the user roles. Interviews were clustered and ranked through a multi-LLM workflow I run myself; synthesis that would have taken weeks finished in days, without losing what people actually said. The output became a product specification written to the domain's real complexity: membership tiers, data compliance that differs by jurisdiction, and a member record that has to survive multiple sites.

The requirements fed a prioritized backlog running to roughly 170 user stories across the platform, the customer-relationship system at its core, each story ranked so the build could be sequenced by value. Against that backlog I ran a systematic scan of more than fifty-five vendors across the stack, and drew the customer-relationship shortlist from it, scoring each candidate on fit, cost band, and reputation, including the ones worth avoiding for reliability or security reasons.

Validated before recommended

The recommendation went back quickly and deliberately: requirements walked through with the roles who gave them, the ranking tested against what the floor actually needed first, and the shortlist stress-tested on cost and reliability before anything was signed. Corrections were made while they were still cheap.

What this shows

Discovery is the product: a fragmented operation taken to a single defensible decision, validated with the people who have to run it.

One of a set of engagement and systems case studies. If a problem like this is on your desk, start with a conversation.