A founder is on a reference call she has been dreading. She has thirty-five minutes. The reference contact is being polite. She is asking the questions she has always asked: were you satisfied with the consultant’s work, would you recommend them. The contact is answering the way contacts always answer, with warm generality and an enthusiasm that does not quite ring true. She is taking notes that she knows will not influence her decision. Both of them know it. They are running through the social ritual of the reference call, and producing none of the data the call is supposed to produce.
This is the most common pattern in SME consultant selection. The reference call is one of the highest-leverage tools available, and almost everyone uses it badly. The fix is structural. A nine-question framework, ordered correctly, produces signal where the default produces noise.
Why the default reference call produces compliance, not data
The structure of the typical reference call is stacked against you. The consultant chose this contact precisely because they expect a positive answer. The contact is presumably a successful client, is grateful for the completed work, and does not want to damage the consultant’s reputation by speaking critically. The default questions, were you satisfied, would you recommend, do not create any space for honest disclosure of challenges. The result is that the answers come back affirmative regardless of the underlying truth.
Reading the social dynamic alone, the call should be re-engineered. The questions you ask should be designed to disrupt the boilerplate, not to invite it. Some questions surface comparability, so you know whether this client is actually like you. Some surface engagement structure, so you understand whether the consultant kept scope discipline. Some surface specific outcomes. The most important one is designed to break the pattern explicitly.
The nine-question structure that produces signal
The nine questions follow a deliberate order. Each builds on the previous, and the sequence matters as much as the wording.
The first three questions establish comparability and scope. Ask about team size, revenue, and prior AI experience at the time of engagement, so you know whether the patterns that worked here are likely to apply to you. Ask about initial scope and whether scope changed, so you understand whether the consultant has discipline around scope management. A reference whose engagement ran on time, within scope, with documented changes, is signalling a consultant with real project discipline.
The middle three questions surface specific outcomes and change management. Ask what specific business outcomes the consultant’s work enabled, then push for quantification. If the answer is “the consultant delivered a chatbot”, ask whether the chatbot improved customer satisfaction, by how much, and how it was measured. Ask whether the team adopted the solution, whether adoption took the form the consultant predicted, and whether the solution worked the way the consultant described. Ask whether, when the engagement ended, the team could maintain and modify the solution on their own, or whether ongoing consultant support was necessary.
The last three questions assess people, structure, and the meta. Ask who the contact worked with most frequently, and whether that person stayed with the project from beginning to end. Ask for an honest assessment of communication style. Ask, on a scale of one to ten, how likely they would be to engage this consultant again for a similar project, and what would move that number.
These nine questions, asked plainly and without padding, produce a richer picture than any reference call where the dominant questions are satisfaction and recommendation.
The disrupter question that breaks the pattern
The single most important question on the call is the sixth one. Ask: what didn’t go well, and what would you do differently if you were to engage this consultant again?
The phrasing matters. It does not ask if anything went wrong. It assumes something did, because something almost always does. The question creates explicit permission for the contact to be honest about challenges. It removes the social pressure to deliver only positive feedback. The contact can answer it without feeling they are betraying the consultant, because the question was framed in a way that invited the answer.
The quality of the response to this question often reveals more than the rest of the call combined. A contact who has a substantive answer, named specifically and discussed openly, is giving you data. A contact who insists nothing went wrong is giving you either an unusually polished engagement or a reference who is too loyal to be useful, and either way you have learned something.
What to listen for that the contact does not say
The most underrated skill on a reference call is listening for what is not said. A contact who enthusiastically explains the value of the consultant’s work but provides vague rather than specific outcomes is signalling that the outcomes are less concrete than the relationship. A contact who mentions, in passing, that adoption took longer than expected or that another consultant was brought in afterwards to refine things is telling you something important.
Reference calls have rhythm. When a contact rushes past a particular topic, that is a signal. When a contact emphasises one dimension warmly but goes vague on another, that is a signal. When a contact mentions a transition in the engagement team, that is a signal. The pattern and coherence of the narrative carries information as much as any individual answer does.
Take notes on what you hear and what you do not hear. The latter is often where the truth lives.
Ask for the reference that did not go perfectly
There is one final question worth asking the consultant directly, before the reference calls happen. Ask: do you have a reference from an engagement where something did not go according to plan initially, and how that engagement adapted? A consultant who can produce this reference is signalling real-world maturity. A consultant who cannot is signalling that they only have frictionless engagements to point at, which is statistically unlikely to be true across a real consulting practice.
The reference from the imperfect engagement is often the most useful call you make. The consultant who handled that engagement professionally, who adapted, who kept the relationship intact, is more predictive of how they will handle yours than the consultant whose portfolio looks all-positive on paper.
If you are about to run a reference call and want help shaping the questions for your specific situation, book a conversation.
