The Decision Map: Who Really Decides to Buy an ERP System

The most common mistake in ERP selling is not a bad pitch. It is an excellent pitch delivered to the wrong person, and the seller can spend months believing the deal is progressing because that wrong person is sincerely enthusiastic.

The number that explains why everything is slow

Gartner’s research on The B2B Buying Journey describes today’s buying group as five to sixteen people spread across as many as four functions. Earlier CEB research, published in The Challenger Customer in 2015 and built on a survey of 3,000 B2B stakeholders, put the average at 5.4 people formally involved in a single decision.

The number that matters more than the count is its effect. That same research found that moving from one decision maker to just two drops the likelihood of completing a purchase from 81% to 55%, and that the likelihood falls into the mid-30s once six people are involved.

Read it correctly: each additional person in the room does not raise your odds — it raises the odds that nothing gets bought at all. Again, the competitor is not a rival vendor. The competitor is the room failing to converge.

Gartner completes the picture with a third figure published in May 2025, from a survey of 632 buyers conducted between August and September 2024: 74% of buying teams exhibited “unhealthy conflict” while deciding, and teams that reached consensus were 2.5 times more likely to describe the deal as high quality.

You are not in the room

The figure that finishes the argument, also from Gartner: when a company works through a purchase, it spends only about 17% of its time meeting all potential suppliers combined. When comparing more than one vendor, any single rep’s share may be 5% to 6%.

94% of your decision is made in rooms you never enter.

This is the central professional fact of the job, and everything after it is built on it: your task is not to convince the person you can see. It is to arm him to convince the people you cannot. The difference between sellers in this market nearly reduces to who has understood that sentence.

The five voices

The names are my own shorthand; the functions are real.

1. The pain owner. The person who lives the problem daily: the warehouse manager counting by hand, the accountant who closes the month in two weeks. His enthusiasm is the most genuine you will meet, and his budget authority is usually zero. Your likely error: mistaking his sincerity for his authority.

2. The economic signer. The founder, the CFO, the CEO. He does not ask about features; he asks about the number, the duration and the risk. You will see him once or twice in the whole deal, and it never closes without him.

3. The technical gatekeeper. IT. He cannot say yes — and he alone can say no. His questions are hosting, integration, backup, and getting the data back out. He is the cheapest voice in the committee if you win him early and the most expensive if you ignore him.

4. The user whose job changes. The accountant, the salesman, the storekeeper. He will not sign and will rarely be asked — but he can make the project fail after the sale, and the reputation of that failure comes back to you on the next deal in the same city. That is the link between selling and why ERP projects fail.

5. The silent loser. The voice nobody mentions. Someone whose standing comes from being the only holder of a piece of information: the person who builds the monthly report by hand and is therefore waited for, or who knows supplier prices nobody else knows. A system that makes information available to everyone takes something real away from him, and he will resist in impeccably professional language: the timing isn’t right, the priority is elsewhere, the system doesn’t suit how we work.

The story: the objection from outside the list

The story is composited from real events, and every identifying detail has been changed.

A mid-sized contractor. The event was clear, the CFO was an advocate, the technical gatekeeper had been satisfied in an early meeting, and the technical proposal had survived comparison against two other vendors. In my own reckoning the deal was ninety per cent done.

Then everything stopped for three weeks.

The source who eventually told me what had happened was none of the three. The objection had come from a purchasing manager who had attended no meeting at all, and he had put it to the founder in one sentence: this system will route every purchase order through three approvals. We work fast, and it will stop us.

The sentence was entirely correct. A disciplined procurement cycle does exactly that. The man had not lied and had not schemed; he had said what he knew about his own job. I had left him off my map because his authority was not in the org chart — it was in twelve years of the founder trusting him.

What saved the deal was not a rebuttal. It was a meeting I asked for with him alone, out of which came an approval matrix keeping purchase orders under a threshold at a single approval. A small design concession, costing nothing, that bought a voice which would otherwise have killed the project for free.

The lesson: a map that contains nobody who loses something is not a map. It is a list of friends.

How the map is actually drawn

Not by asking “who decides?” — the answer to that is always a courtesy. It is drawn by asking about mechanism:

“Last time you bought something this size, who did the decision pass through before it was signed?”

“Who inside the company might think this is the wrong time?”

“If everyone agreed today, what is the next administrative step exactly?”

The second is the most valuable and is almost never asked, because it looks like inviting trouble. It is the opposite: an honest answer hands you the silent loser’s name before he kills your deal, and an answer of “nobody” tells you your contact either does not know his company or does not trust you yet.

The third is a seriousness test with no confrontation: a person who cannot name the next administrative step is not as close to buying as he appears.

Leave with one page

Every deal should produce one page carrying, for each name: what he measures you on, what he loses, and whether he can say no alone. Three columns, no more.

If the second column is empty against every name, you have not drawn the map. You have written an attendance sheet.

Checklist before you move on

• How many names are on your map? If it is one, it is not a map.
• Who on the list can say no alone?
• Who loses something if the project succeeds — and have you spoken to him?
• Do you know the next administrative step after technical approval, by name?
• What have you armed your contact with to defend you in your absence — a file? a number? one sentence he can repeat?

Next station

You know who is in the room. But the room itself differs so fundamentally by company size that these are two markets, not one: mid-market versus enterprise.