Start a learning analytics discussion with governance

Learning analytics deals fall apart in a predictable way. A provost hears the word predict and hears a promise. Faculty hear it and hear surveillance. Your job as the account executive is not to defend the algorithm. It is to slow the conversation down to a single question, the data behind it, and the people who get to govern how it is used. Get that framing right early and the evaluation becomes something a governance committee can actually approve.

Why this conversation goes wrong

Most learning analytics pitches start with the dashboard. That is the mistake. The moment a prospect sees a risk score next to a student's name, they stop thinking about the underlying data and start thinking about what happens to that student. A provost asks if the tool can predict failure. A faculty senate asks about bias. A privacy officer asks who sees the output. If you answer with confidence about the model, you sound like you are promising certainty about individual students, and no responsible institution will buy that promise.

The deeper problem is that most institutions have gaps in their data. Adjunct sections, dual enrollment programs, and partner campuses often feed into the student information system late or not at all. A model built on incomplete data will quietly favor the students who show up cleanly in the system and miss the ones who do not. If you do not name that gap, the buyer will find it during a faculty review, usually after they have already spent months on your evaluation.

How to prepare

Before you sit down with a provost or a data governance committee, get specific about three things.

1. The decision. Ask internally and out loud what decision the analytics would inform. An early alert to an advisor is a different product than a course placement recommendation. A retention forecast for a budget meeting is different again. Vague talk about insights invites vague scrutiny. A named decision invites a scoped review.

2. The data. Find out which student populations and course types are represented in the data your buyer would feed into the system. Ask specifically about adjunct led sections, night and weekend programs, and any partner or dual enrollment arrangements. These are the places where data usually goes missing, and they are also the places where equity concerns usually start.

3. The reviewers. Learning analytics touches privacy, faculty governance, and institutional research at once. Find out who sits on each of those committees before your first substantive meeting. If you wait for the provost to loop them in, you lose control of the timeline and the framing.

Read through the Education practice conversations hub before a first call like this. Several of the scenarios there, including the one on how to Guide a student considering leave after a family crisisshare the same underlying skill. You are not selling certainty. You are helping an institution make a bounded decision with the information it actually has.

What to say

Here is how a conversation like this can go. A provost named Dr. Okafor opens with the question you should expect.

Okafor says, "We keep hearing about predictive analytics. Can your system tell us which students are going to fail before it happens. Our faculty senate is already worried about bias, so I need a real answer, not a sales pitch."

The account executive, Priya, does not defend the model. She redirects to the decision. Priya says, "The system surfaces patterns in the data your institution provides, and your team decides what to do with those patterns. It does not make predictions about individual students on its own. Before we talk about accuracy or bias, I want to know what decision this would actually support. Would an alert go to an advisor for outreach, or would it factor into something with more weight, like course placement."

Okafor says the intent is advisor outreach, nothing tied to grades or academic standing. Priya asks the next question. She says, "That is a good scope to start with. Now tell me about your data. Are adjunct led sections and any dual enrollment programs feeding into your student information system the same way your full time faculty sections do."

Okafor pauses, then admits that adjunct sections often lag behind by a term or more, and some dual enrollment data comes through a separate partner system that does not sync automatically.

Priya says, "That gap has to be visible in any evaluation we run, because those are often the students most likely to need outreach and least likely to be represented in the data. I would like to bring institutional research, your privacy office, and someone from faculty senate into the same room to define one limited use case, advisor outreach alerts for a single term, and agree on the review criteria before we build anything. In that meeting we can walk through exactly which inputs feed an alert and how each one gets documented."

Notice what Priya never does. She never claims the tool can identify at risk students with confidence. She never treats the data gap as a footnote. She puts governance inside the evaluation itself, as a working session, not as a signature required at the end.

Mistakes to avoid

Do not answer the bias question with a feature list. Naming an audit log or a fairness dashboard does not answer a faculty senate's real concern, which is whether the tool will be used to make judgments about students based on incomplete information. Answer the concern, not the checkbox.

Do not let the provost define the use case alone. A provost focused on retention numbers may describe a use case that faculty would reject outright, like flagging students for probation based on an early alert score. If you build a proposal around that framing, you inherit the fight when faculty senate finds out.

Do not treat the missing adjunct or dual enrollment data as a detail to mention once and move past. Write it into the evaluation plan. If the institution decides to proceed without that data, that should be a documented decision, not a gap nobody noticed.

Do not skip straight to a pilot. A pilot with the wrong scope or the wrong reviewers in the room just delays the real objections instead of resolving them.

How to practice and score this

Take an actual dashboard mockup or product one pager and write down, in one sentence, the specific decision it is meant to support. If you cannot write that sentence cleanly, you are not ready for the meeting.

Ask a colleague to play the skeptical provost or a faculty senate member and to name one data source your institution's system probably does not capture well. Adjunct sections are the easiest example, but partner campuses and continuing education programs work too.

Run the exchange as a roleplay. Score yourself on three things. Did you name the decision before talking about the model. Did you ask about the data gap before the other person raised it. Did you propose a specific governance meeting with named roles, rather than a vague promise to loop people in later. A manager reviewing the roleplay should listen for whether you ever describe the tool as making a decision, rather than surfacing information a human reviews. DealSpeak scenarios built around this kind of conversation score exactly those moments, so run it a few times before the real meeting. If your team is also selling into library systems, Explore a library credential partnership covers a related governance conversation worth the same kind of rehearsal.

Institutions do not reject learning analytics because the technology is weak. They reject it when a vendor cannot answer a straightforward question about what decision the tool supports and who is accountable for it. Ask about the decision first, name the data gaps before anyone else does, and bring governance into the room early. That is what turns a nervous provost into a sponsor.