Why the standard script breaks
Open almost any discovery script and it reads as a sequence. Build rapport, set the agenda, establish current state, quantify the pain, identify the decision process, agree next steps. The content is usually fine. The problem is the form. A sequence assumes step four happens after step three, and the first real buyer breaks that assumption in the first two minutes by asking about price, or by opening with a complaint about their current vendor.
At that moment the rep is holding a document that no longer describes the call they are on. What follows is one of two failures. Either they abandon the script entirely and improvise for the rest of the hour, or they keep trying to steer back to step two while the buyer is already at step five, which reads as not listening.
The fix is not a better sequence. It is to stop thinking in sequence at all. A discovery call has a small number of things that must be true when it ends. Their order is negotiable. Their presence is not.
Three load-bearing parts
- An agenda. Thirty seconds at the top: what you would like to cover, how long you have, and what a useful outcome looks like for both sides. The point is not politeness. The point is a shared reference you can return to later without it feeling like you are pulling rank. "You mentioned you wanted to leave with a sense of cost, can I ask two things first" only works if the agenda was stated.
- A diagnosis. Everyone claims to do this part. The test is not how many questions were asked. It is whether you can state the buyer's situation back in their own words, including the part they are not proud of, and have them agree with it. If your notes are all facts and no consequences, you ran a data collection exercise, not a diagnosis.
- An agreed next step. Not "I will send something over". A named action, a named owner, a date, and the buyer saying yes to it out loud while you are both still on the call. This is the cheapest of the three to get and the one most often skipped, because it lives in the last minutes, which are the minutes that get eaten.
Three is roughly the limit of what a person can hold while also listening. That constraint is the whole design. A structure you have to consult is not a structure you have during a conversation.
An opening that survives the first question
| Beat | What it does | Roughly |
|---|---|---|
| Reason for the call, one sentence | Gives the buyer something concrete to react to | Ten seconds |
| Agenda plus the time box | Makes the shape explicit while it is still cheap to agree to | Twenty seconds |
| Permission to ask questions | Turns an interrogation into an invitation | Five seconds |
| The first real question | Hands the floor over before the buyer takes it | Then stop talking |
Notice what is missing: there is no pitch in the opening. The most common way an opening collapses is a rep who feels they have to earn the meeting, spends a minute and a half on the company and the category, and hands the buyer a reason to go defensive. The buyer's first interruption then arrives as a challenge instead of a question, and the call starts from a worse place than silence would have.
A more formal version of this same opening has a name. The Sandler up-front contract makes the agenda and the permission ask an explicit agreement about how the call is allowed to end. The structure is the same one described here; the contract is a way of saying it out loud.
What to do when it breaks anyway
- The buyer opens with a question. Answer it in one sentence, then say what you would need to know to answer it properly. You have now done the agenda and the permission ask in the same breath, without announcing either.
- The call is half the length you booked. Drop the agenda recap and drop the demo. Keep the diagnosis and the next step. Two of the three is still a call worth having, and the next step is never the one you drop.
- You never got a straight answer about the problem. Say so before you close, in plain words: "I am not sure I understand yet what makes this worth changing this quarter." Buyers usually answer that honestly. If they do not, that is also an answer.
- The buyer wants a proposal now. A proposal built on scope nobody established is a document you will spend the next two weeks defending. Offer the next step instead, and be specific about what it will make possible.
A good conversation with a vague ending is a call you will have again from the beginning. Reserve the closing minutes before the call starts, not when the clock is already gone.
The same structure, very different words
None of this is specific to enterprise software, but the wording is. A property owner picking up a call from an acquisitions rep will not sit through an agenda that sounds like a business meeting. The three parts still apply, compressed into a sentence or two: "I buy houses in the area. I would like to ask a few things about the property and your timeline, and if it makes sense I will tell you what I can do. Five minutes, is that alright?" That is an agenda, a time box and a permission ask, said the way a person talks.
Diagnosis in that call means condition, occupancy, title, timeline, and the reason they are considering selling at all. The last one is the one reps skip, because asking why somebody is selling their house feels intrusive. It is also the only one that tells you whether the price conversation is going to be possible.
And the next step is a walkthrough with a date on it, not "I will run some numbers and get back to you". The same rule as everywhere else: an action, an owner, a date, agreed out loud. A seller who says yes to a Thursday walkthrough is a different situation from a seller who says yes to being called again sometime.
Why a discovery call structure gets roleplayed, not memorized
Three parts fit on a sticky note. That is deliberate, and it is still not enough. The opening in particular has to be said out loud enough times that it comes out at conversation speed rather than reading speed, because an opener that sounds memorized is worse than no opener at all. The buyer hears the seam, and the seam tells them this is a process being run on them.
The real thing to role play is not the opening though. It is the recovery. The measure of whether a rep has this structure is how fast they get back to it after the call goes somewhere unplanned, and that speed cannot be read off a page. It comes from being knocked off the track many times, by an AI buyer who does it differently on every run, until finding the way back stops requiring thought.
That is what a roleplay is for. In pichi.ai a rep runs the call by voice against an AI buyer who interrupts, jumps ahead, and does it differently on the next run, then gets scored on the three things that were supposed to be true at the end. The first seconds of the call are their own exercise, and we wrote about the cold call opener separately.