Customer reference calls are one of the few diligence steps every buyer does and almost nobody structures well. The typical version: the seller provides a list of five or six accounts, a buyer books thirty minutes with each, asks some version of "how's the product, would you recommend it," gets some version of "yeah, it's good, we're happy," and moves on having confirmed almost nothing that wasn't already implied by the customer's willingness to take the call in the first place. That's not a criticism of the buyers running these calls; it's what happens by default when a genuinely useful diligence step gets treated as a formality to check off rather than a source of real signal.

The fix isn't more calls. It's a different structure for the calls already happening, built around a simple idea: the seller-selected reference list and the polite, product-focused question set are both optimized to produce reassurance, not information. Getting real signal means deliberately working against both.

Why the standard reference call fails, by design rather than by accident

Start with the list. A seller preparing a sale process is not going to hand a buyer their most disgruntled customer's phone number, and that's not dishonesty, it's a rational response to an adversarial-ish situation where the seller's advisor has explicitly coached the customer base on messaging before any call happens. The customers who end up on a reference list skew toward long tenure, genuine satisfaction, and, often, a personal relationship with the founder that makes them inclined to be generous on a call they know matters to someone they like.

Then add the call dynamic itself. Most customers, even genuinely satisfied ones, are reluctant to be blunt with a stranger about a vendor relationship on a recorded or semi-formal call, especially when they don't know who's calling or exactly why, and especially when the honest answer involves some nuance ("it's fine but we've looked at alternatives twice") that's more effort to explain than "yeah, we're happy." The result is a call that confirms the CIM without testing it, run with a customer selected for their likelihood to confirm it in the first place.

Fix 1: Don't rely only on the seller's list

Ask for the complete customer list, with revenue and tenure by account, not just the curated reference names, and pick two or three calls yourself from outside the provided list, ideally including at least one account that's mid-tenure rather than a ten-year relationship, and, if any recently churned customers are willing to talk, one of those. A seller resisting a reasonable request for the full account list, or resisting off-list calls entirely, is itself informative; a seller with nothing to hide about customer sentiment across the base generally has little reason to gatekeep who a buyer talks to that tightly.

A departed customer is worth pursuing even when it's uncomfortable to ask for that contact, because a former customer has no continuing relationship to protect and will typically be considerably more candid about why they actually left than any current customer will be about anything they're unhappy with.

Fix 2: Ask about the last renewal, not the product

The single most predictive question available on a reference call isn't "how do you like the product." It's some version of: "walk me through your last renewal conversation, what happened, who was involved, did you look at anything else." This works because a renewal is a real event with real details a customer either remembers accurately or doesn't, which makes it much harder to give a vague, pleasant non-answer to than a general satisfaction question.

The answer reveals several things at once: whether the renewal was a genuine internal decision process or a passive auto-renewal nobody thought hard about, whether the customer shopped competitive alternatives before renewing (and if so, which ones, which is itself useful competitive intelligence), whether pricing came up as a point of friction, and who internally championed staying with the vendor, which speaks directly to how dependent the account is on one specific relationship rather than genuine organizational buy-in.

Fix 3: Ask what would actually make them leave

This question feels aggressive to ask on a first reference call, which is exactly why almost nobody asks it directly, and exactly why it produces real signal when asked plainly and without apology: "what would have to happen for you to switch to a different vendor?" Most customers, even happy ones, will actually answer this candidly, because it's phrased as a hypothetical rather than a criticism, and because most people find it easier to describe a condition for leaving than to volunteer unprompted complaints about a current vendor.

The answers cluster into a few recognizable categories, each with a different implication. "Nothing, honestly, we're locked in operationally" is the strongest possible switching-cost signal available from a customer conversation. "If pricing went up a lot" identifies a specific, testable risk. "If a competitor built feature X" names a concrete product gap worth checking against the company's actual roadmap. "We've actually looked before and decided against it" is worth following with an immediate follow-up: looked at what, and why did you stay.

Fix 4: Test the switching-cost claim directly, not just its symptoms

Related to the point above, but worth asking as its own explicit question rather than assuming it's covered: "if a competitor called you tomorrow with a product that did roughly the same thing for 20% less, what happens?" This is a direct, concrete stress test of whatever switching-cost story is being told elsewhere in diligence (a fuller methodology for testing that claim independently is covered in a companion piece on this site), and a customer's actual answer, including how quickly and how confidently they answer it, is more informative than any retention percentage on a summary slide.

A customer who has to think about it for a while, or who says something like "honestly, if the switching cost weren't so high we probably would have looked already," is telling you the retention number is doing more work than the product relationship is.

Fix 5: Listen to the hedge, not just the words

Reference calls carry as much signal in how something is said as in what's said, and it's worth explicitly listening for a few specific patterns rather than only transcribing the literal answers:

  • The long pause before "we're happy." A genuinely satisfied customer usually answers a satisfaction question quickly and specifically. A hesitation before a positive answer is worth a direct, gentle follow-up: "you paused there, what's the honest version?"
  • Qualified praise. "It does what we need it to do" is a materially weaker statement than "we couldn't run our operation without it," even though both sound positive read off a transcript.
  • Enthusiasm about the product but vagueness about the vendor relationship. A customer who lights up describing a specific feature but goes flat describing account support or responsiveness is telling you where the actual risk sits.
  • Anyone who brings up price unprompted. If pricing comes up without being asked about, it's on their mind more than a scripted answer to "how's the product" would ever reveal.

A call structure that puts this into practice

A simple structure for a thirty-minute reference call that fits all of the above without turning into an interrogation:

  1. Open with context, briefly. Who you are, why you're calling, and an explicit, genuine assurance that candor is more useful to you than reassurance. This framing alone measurably changes how customers answer the rest of the call.
  2. Ask the renewal question first, while the customer is still warmed up and before any pattern of polite, general answers has been established by earlier questions.
  3. Ask the switching-cost stress test ("a competitor calls tomorrow, 20% cheaper, what happens") directly, not as a follow-up buried later in the call.
  4. Ask what would make them leave, framed as a hypothetical, not a complaint-fishing expedition.
  5. Ask about the vendor relationship specifically, separate from the product: responsiveness, account management, how issues get resolved when something goes wrong.
  6. Close by asking who else at their organization would have a different, useful perspective, which occasionally surfaces an internal skeptic worth a follow-up call the seller's reference list was never going to offer voluntarily.

None of this requires more calls than a standard reference process already involves, and none of it requires an adversarial tone with a customer, most of the sharpest answers above come from customers who are perfectly happy but were simply asked a question specific enough to require a real answer instead of a polite one. The reference call is one of the cheapest diligence steps available and, structured this way, one of the highest-signal, precisely because it's the one conversation in the entire process with someone who has no financial stake in how the deal turns out.