Your reference program isn't a program. It's a favor economy.

Let me describe a reference program I've seen at dozens of B2B SaaS companies.

Sales needs a reference call. Someone pings customer marketing. Customer marketing pings the CSM. The CSM has three names they always use — the same three customers who've said yes before. One of them agrees, again. The call happens. No one tracks how many times that customer has been asked. No one asks what she gets out of it.

Most teams don't measure this, but they should.

If you're asking the same customer more than 1–2 times per quarter, you don't have a program. You have a favor economy held together by the goodwill of a few customers who happen to like you.

And it breaks exactly when you need it most. A new enterprise deal requires a reference from a specific vertical. A key account gets asked for the fifth time this quarter and finally says no. The CSM who knew all the advocates leaves, and suddenly no one knows who to call.

The companies that don't have this problem built something different. Not a bigger list — a system.

The mistake everyone makes first

Reference programs fail before they launch because they get assigned to the wrong owner with the wrong mandate.

When Sales owns it, it becomes transactional: "Who can we get on a call right now?" Not "Who is right for this call?" or "How can we build a values-based exchange?"

When Marketing owns it as proof, it becomes static: "Let's turn this into a case study." But nobody is actively managing reference relationships or connecting program activity to pipeline.

Both are directionally right. Both miss the point.

"A reference program is relationship infrastructure. It sits at the intersection of CS, sales, and marketing — and it requires someone who can operate in all three."

A reference program needs to: build supply over time, activate on demand, and connect directly to pipeline. That's customer marketing. If that function doesn't exist or doesn't own it, the program will always revert back to the favor economy under pressure.

Get the ownership question right before you build anything else.

"Not public" doesn't mean "not useful"

The customer I spent two years trying to unlock was an enterprise account at Invoca. Locked down by PR and legal. Not allowed to participate in anything requiring public attribution. Every time we asked, the answer was some version of: they love us, but they can't say so publicly.

Most programs would have written them off. We didn't. Because advocacy isn't binary. It's a spectrum.

That customer never showed up on our website. But they influenced deals, analysts, and product direction.

If you only measure Tier 1, you're ignoring most of your usable advocacy. Don't treat "not public" as "not useful." Build the full spectrum into your program from day one.

What advocates actually want

Most reference programs are designed around what the company needs. The ones that last are designed around what the advocate gets.

This isn't a feel-good principle — it's a retention problem. An advocate who feels used stops saying yes. An advocate who feels genuinely valued shows up well, stays engaged, and proactively brings you new opportunities.

The motivations vary. Some advocates want peer visibility: speaking slots, co-authored content, features that put their name somewhere their boss will see. Some want influence and early access: product roadmap input, beta features, a direct line to your team. Some just want reciprocity: a direct note saying a deal closed because of their call. A handwritten card. An acknowledgment that their time mattered.

Know which type your advocates are. Offering a gift card to someone who wanted a speaking slot — or a speaking slot to someone who just wanted to feel seen — is worse than offering nothing. It signals you don't actually know them.

Maria Ogneva is co-founder of Rally, a B2B SaaS customer marketing consultancy. Rally builds customer marketing programs as connected systems, tied to the metrics that matter to revenue leadership.