I Ran 4 Customer Advisory Boards. 3 Turned Into Wishlist Sessions. Here's What I Changed.

The first customer advisory board I organized at Finvestfx felt like a win just because people showed up. We had 6 enterprise clients in a room, I had a deck with product updates, and I was ready to "gather feedback."

What I got was a prioritized list of 23 feature requests, zero strategic insight, and a nagging feeling that I'd just turned our biggest advocates into a glorified support ticket system.

I've run CABs at three companies now. The early ones were useless. The recent ones at Sonic Linker actually shaped our roadmap. Here's what changed.

Stop asking "what features do you want" and start asking "what are you trying to do"

The problem with most CABs is that we treat customers like product designers. We ask what they want. They tell us. We write it down. Everyone leaves feeling productive, and nothing useful happened.

At Finvestfx, I was managing 20+ treasury and forex clients. In my second CAB, I changed the framing completely. Instead of "what should we build next," I asked: "walk me through the last time you had to reconcile a forex transaction manually."

One client described a 40-minute process involving three spreadsheets, two phone calls, and a prayer that the timestamps matched. Another said they just gave up and accepted the discrepancy if it was under $500.

That's when I realized we didn't need the 5 features they'd requested. We needed one workflow that removed the phone calls entirely. We built it in 6 weeks. Retention on that client segment jumped from 71% to 94% over the next quarter.

The shift was simple. I stopped asking for solutions and started asking for situations. Real ones. Recent ones. Messy ones.

Invite the right mix, or you'll just hear the loudest voice

At Sonic Linker, we were building an AI platform for GTM teams. Our first CAB had 8 people. 7 were power users who loved the product. 1 was a skeptic who barely used it.

Guess who drove the conversation? The skeptic. And not in a useful way. He wanted features that would make the product feel more like the old tool he was comfortable with. The power users nodded along because they didn't want to seem difficult.

For the next session, I changed the mix. 3 power users, 2 moderate users, 1 brand new customer, and 2 people from segments we wanted to expand into but hadn't nailed yet. I also set a rule: no one could suggest a feature without first describing a specific problem they'd encountered in the last two weeks.

That constraint alone killed half the wishlist thinking. When people had to ground requests in recent pain, the conversation got real fast. The new customer admitted she didn't understand our onboarding. The power users realized they were working around bugs we didn't even know existed. The folks from new segments told us our positioning was confusing.

None of that would've surfaced if I'd just asked "what do you want us to build?"

Use the CAB to validate, not generate

This is the biggest shift I made between Finvestfx and Sonic Linker. I stopped treating the CAB as a discovery session. It's a validation session.

By the time we got to our third CAB at Sonic, I already had a rough thesis about where the product needed to go. We'd done 30+ user interviews, analyzed drop-off points, and had a hypothesis: our AI outputs were good, but people didn't trust them enough to act on them without manual review.

I didn't ask the CAB "should we build trust features?" I showed them three rough concepts and asked: "if we shipped this next month, would it change how you use the product?" Then I asked them to be honest about which one they'd actually use, not which one sounded cool.

Two of the three concepts got polite interest. One got "holy shit, I'd use that today." We built that one. Shipped it in 5 weeks. Adoption hit 60% in the first month.

The CAB didn't tell us what to build. It told us which bet to take. That's way more valuable.

The structure that actually works

Here's the format I use now:

Pre-meeting: Send a one-pager with 2-3 focus areas. Ask attendees to come with a recent example related to each area. Not hypotheticals. Actual stories from the last 30 days.

First 30 minutes: Go around the room. Everyone shares one recent pain point. No solutions yet. Just problems. I take notes, probe for details, and shut down anyone who jumps to "you should build X."

Next 45 minutes: I present 2-3 things we're considering. Rough mocks, not polished decks. I ask: would this solve the problem you just described? Would you actually use it? What would stop you?

Last 15 minutes: Open discussion, but I'm steering it toward tradeoffs. If we do A, we can't do B for 3 months. Which one matters more right now?

Post-meeting: I send a summary within 24 hours. Three things we heard, two things we're doing because of it, one thing we're explicitly not doing and why.

The real unlock

The best CABs I've run didn't feel like product feedback sessions. They felt like strategy conversations. Customers weren't telling me what to build. They were telling me what's actually hard about their jobs, and I was translating that into product decisions.

That translation is my job, not theirs. The moment I stopped outsourcing roadmap prioritization to a room of people who don't build software, the CAB became useful.

Now it's one of the highest-signal inputs I have. But only because I stopped asking customers to do my job for me.