SPIN Selling for PEO Reps, Powered by AI

SPIN Selling is almost forty years old, and it is still the best framework for a PEO discovery call, because PEO is a considered purchase that solves problems the buyer has often stopped noticing. Neil Rackham's research, published in 1988, found that in larger sales the reps who won asked more questions of a particular kind, in a particular order, and closed less. That matches two decades of our own experience selling PEO. The rep who walks in with a rate sheet loses to the rep who walks in with questions.

The problem with SPIN in practice is that generic questions get generic answers. "What challenges are you facing with HR?" is a Problem question in form only. The questions that work are specific to the prospect, and specificity requires preparation that most reps do not have time for. That is the part AI can take over, and the part we built into HALO IQ.

The four question types, mapped to a PEO deal

Situation questions establish facts. In PEO discovery they are about headcount, states, payroll provider, current WC carrier and program, benefits in place, who handles HR today. They are necessary and the buyer finds them tedious, so a good rep asks as few as possible and gets the rest from the submission.

Problem questions surface dissatisfaction. Where does payroll go wrong. How long did the last WC audit take. What happened the last time an employee filed a complaint. Who answers the phone when a manager has a termination question on a Friday afternoon.

Implication questions are where PEO deals are won, and where most reps stop short. They connect a problem to its consequences. If the WC audit found misclassified payroll, what did that cost at the premium level, and what does it mean for the mod next year. If the handbook has not been updated since the last state law change, what happens in the next wage-and-hour claim. If the owner spends a day a week on HR, what is not getting sold.

Need-payoff questions let the buyer state the value of a solution in their own words. If the mod came down and the audit took an hour instead of a week, what would that be worth. If every manager had an HR person to call, what would change.

The sequence matters. Implication questions before Problem questions feel like an interrogation. Need-payoff before Implication produces a lukewarm answer because the buyer has not yet felt the size of the problem.

Why the submission is the best discovery prep a rep will ever get

A rep preparing for a PEO discovery call traditionally has a company name, a headcount, and whatever the broker said on the phone. A rep working in HALO IQ has, by the time the call is scheduled, the parsed ACORD 130, the loss runs aggregated by policy year, the mod projection from ModIQ, the class code mix, the public enforcement record, and often a census.

That is a complete picture of the prospect's Situation before anyone has asked a Situation question. More importantly, it contains the Problems. A loss run with five claims in a single class code and a long report lag is a Problem waiting to be asked about. An ACORD 130 that lists three entities with the same address is a Problem. A mod that went from 0.85 to 1.20 over three years is a Problem, and the Implication is already quantifiable from the premium.

This is what SPIN content generation in HALO IQ does. It reads the deal context, identifies the specific conditions in the submission that a skilled rep would ask about, and drafts the questions, in sequence, tied to the evidence.

What a generated SPIN set looks like

Here is a hypothetical, built from a made-up submission to show the shape of the output rather than a real deal. Take a 140-employee food processing plant in the Central Valley, two shifts, with a mod that rose over the last three rating periods, a cluster of strain and sprain claims on the second shift, and a handbook last revised several years ago.

The Situation questions the AI drafts are few, because most of the facts are already in hand: confirm the second shift's supervision structure and whether the plant has a safety committee.

The Problem questions are specific: ask what the plant knows about why second-shift claims run higher than first, and ask how the last Cal/OSHA interaction went. Ask who reviewed the handbook most recently and what prompted it.

The Implication questions tie each problem to money or risk: ask what the mod increase has done to the WC premium over three years, in the prospect's own figures, and ask what a wage-and-hour claim under current California law would look like given the handbook's age. The rep has the PAGA IQ model and the mod projection open while asking, so the prospect's answer can be checked against the numbers in real time.

The Need-payoff questions invite the prospect to describe the plant with a working safety program, a mod trending back down, and a current handbook, and to say what that is worth to them.

Every generated question carries a reference to the data that produced it, so the rep can see that the second-shift question came from the claims by shift and date on the loss run, not from a template.

Where the AI helps and where it does not

The AI is good at the part of preparation that is reading and pattern-matching. It reads the whole submission every time. It does not forget that the loss run showed a claim type the ACORD did not mention. It drafts questions in the right order and ties them to evidence. It does this in seconds, for every deal, which means a rep with twenty open opportunities gets prepared discovery for all of them instead of for the three they had time to study.

HALO IQ runs this generation through the IQ Agent, which means two frontier models draft independently and the output is scored and merged. For discovery questions that matters less than it does for an underwriting opinion, but it does catch the case where one model fixates on a single feature of the submission and misses the rest.

What the AI does not do is ask the question. SPIN works because of what happens after the question: the pause, the follow-up, the willingness to let the prospect sit with an uncomfortable implication. A rep who reads generated questions off a screen in order will get worse results than a rep who internalizes them and then has a conversation. The generated set is the preparation a great rep would have done with an afternoon free. It is not the conversation.

It also does not replace the rep's own knowledge of the prospect's industry. The AI can see that a food processor has strain claims on the second shift. The rep who has walked a dozen packing floors knows to ask about the line speed and the rotation schedule.

How SPIN output flows through the deal

The generated questions live in the deal record in ConnectIQ, alongside the submission documents and the mod projection, so the rep and the underwriter see the same preparation. After the call, the rep's notes on the answers go into the same workspace, and the Implication answers in particular feed two later steps.

The first is the Cost of Inaction analysis, which takes the prospect's own figures from the discovery call, their premium history, their estimate of HR time, their exposure as they described it, and builds the number the proposal will put in front of them. A Cost of Inaction built from the prospect's own answers is credible in a way that a vendor's estimate never is.

The second is Executive Summary AI, which drafts the proposal's narrative from the structured data in the wizard and from the discovery context. A proposal whose executive summary opens with the problem the prospect described in their own words, and then shows the number they agreed to, is a proposal that reads as if it was written for them, because it was.

The point

Rackham's finding was that the best reps in complex sales ask better questions, and that better questions come from preparation and sequence rather than charm. PEO is a complex sale. The preparation has always been available in the submission; nobody had time to do it on every deal. Now the preparation is done before the call is booked, the questions are specific to what the data shows, and the rep's time goes to the conversation, which is the only part that was ever worth their time.