There is a poster on the wall of half the B2B marketing teams I have worked with. It has a stock photo of a smiling woman, a name like "Marketing Mary," and a list of facts. She is 42. She has two kids. She reads Forbes and listens to podcasts on her commute. She values efficiency and hates jargon.
Nobody has ever closed a deal with that poster. No rep has opened it before a call. It was built in a workshop two years ago, printed once, and quietly ignored ever since. When I ask a sales team who their buyer is, they never describe Mary. They describe the VP of Ops who ghosted them last week, the procurement lead who nuked the price, and the junior analyst who actually loved the product but had zero budget authority.
That gap is the whole problem. Most buyer personas are fiction dressed up as research. They describe a person who does not exist, ignore the four other people in the room, and never touch the CRM where deals are actually worked. This post is about building personas that do the opposite: grounded in your own won deals, mapped to the real buying group, and wired into the system your reps use every day.
What a buyer persona actually is
Let me clear up the confusion first, because most teams mix two different things and get neither right.
A buyer persona is a profile of a person. It describes one human in the buying group: their role, the pressure they are under, what they are trying to fix, what makes them say yes, and what makes them kill a deal. It answers the question "who am I talking to and what do they care about."
An ideal customer profile is a profile of a company. It describes the account you want to sell to: industry, headcount, revenue, tech stack, growth stage. It answers "which companies should we go after." I wrote a full breakdown of how to build one from real data in the data-driven ICP guide, and the two documents work as a pair. The ICP tells your team which doors to knock on. The personas tell them who opens each door and what to say once it does.
Get this backwards and you end up with beautiful account targeting and generic messaging, or sharp messaging aimed at the wrong companies. You need both, and they have to agree with each other.
A persona is a person, not a demographic.
The useful part of a persona is not the age or the podcast habit. It is the job they are trying to get done and the thing that makes them walk away. Everything else is filler.
Why most personas are fiction
Three things go wrong, and they compound.
The first is that personas get built from imagination instead of data. A team sits in a room, guesses what the buyer probably thinks, and writes it down with confidence. There is no interview, no call recording, no CRM export. The persona reflects what the sales leader believes about buyers, which is often what was true five years ago in a different market.
The second is the demographic filler. Age, income, favorite media, marital status. In B2B, almost none of it changes how you sell. Whether your economic buyer is 38 or 51 does not move the deal. Whether they get promoted based on cost savings or revenue growth moves everything. Teams pad personas with easy-to-guess demographics and skip the hard, useful part: the buying trigger and the decision criteria.
The third is that personas describe one person when the deal has many. B2B buying is a group activity now. Gartner puts the typical buying group at six to ten people, rising to thirteen on more complex purchases, and Forrester found that deals above a million dollars can pull in fourteen to twenty-three stakeholders. Your "Marketing Mary" persona describes exactly one of them and pretends the rest are noise. They are not noise. They are the reason the deal stalled.
That middle number is the one that keeps me up. Gartner reported in 2025 that 74% of B2B buyer teams show what they call unhealthy conflict during the decision. The people you are selling to do not agree with each other. If your persona work assumes one rational buyer weighing your features, you are modelling a room that does not exist. Real deals die in the argument between the champion who wants you and the finance lead who wants the cheaper option.
Map the buying group, not the buyer
The fix starts with accepting that you are selling to a committee. You do not need a persona for every one of thirteen people. You need one for each distinct role that can advance or block the deal. On most B2B purchases that is four or five.
There is a fifth role that never makes it onto the poster: the blocker. Sometimes it is a person whose team owns a competing tool, sometimes it is the exec who inherited the last failed rollout and will not sit through another. You cannot always win a blocker, but you have to name them, because a deal with an unnamed blocker is a deal you will lose without ever knowing why. I dug into how to work all of these threads at once in the piece on multithreading deals so they stop slipping.
Each of these roles gets a short persona. Not a life story. Half a page: what they are trying to get done, the one metric they answer for, the objection they raise first, and the single line that moves them.
How to build a persona from real data
Here is the method I run with clients. It takes about a week and produces something reps will actually use, because it comes from deals they recognize.
The interviews are where the work lives. You are looking for three things in every conversation. What made this land on their desk in the first place, because that trigger is your prospecting signal. Who else weighed in and what they pushed back on, because that maps the committee. And what nearly killed the deal, because that objection is the thing your persona has to answer before a rep ever gets asked.
Call recordings do half the job for you. If you run Gong or even just record calls in your conferencing tool, the language is already sitting there. When six different economic buyers use the same phrase to describe the problem, that phrase goes in the persona and then in your outbound. This is the same raw material that makes a discovery call framework work, and the two efforts feed each other.
One caution on interviews: talk to people who bought and people who did not. Lost deals tell you what your persona is missing. If every won buyer says "your onboarding sold me" and every lost buyer says "I never understood how you were different," you have found a gap that no internal workshop would have surfaced.
Wire the persona into the CRM or it dies
This is the step almost everyone skips, and it is the reason the poster ends up on the wall instead of in the deal. A persona that lives in a slide deck gets looked at once. A persona that lives in the CRM gets used on every deal.
In HubSpot or Salesforce, add a contact property for buyer role: champion, economic buyer, technical evaluator, end user, blocker. Make reps tag every contact on an open deal. Now your pipeline knows something it never knew before, which roles are engaged on each deal and which are missing. A deal with a champion but no economic buyer tagged is a deal without a path to signature, and your forecast should treat it that way.
Then tie the persona to the message. When a rep adds a technical evaluator to a deal, the next step should be obvious because the persona already told you what that role needs: the security doc, the integration overview, the reference call. You can automate the nudge. We build these routing and reminder flows for clients as part of a CRM and RevOps setup, and the simple version is a workflow that fires a task when a role is engaged, pointing the rep at the right asset for that persona.
Where AI helps and where it lies
I get asked constantly whether you can just ask a model to write your personas. You can, and the output looks great, and that is exactly the trap. A generic model will hand you a confident, well-written persona that is indistinguishable from the fiction you are trying to escape, because it is drawing on the same generic B2B content everyone else trained on. It will invent a plausible economic buyer who has nothing to do with your actual market.
Where AI earns its place is on your own data. Feed a model your real call transcripts and won-deal notes and ask it to cluster the objections, surface the repeated phrases, and flag which roles showed up in your best deals versus your worst. That is genuine pattern-finding on evidence you own. We wire this kind of analysis into client workflows through AI automation, and the rule is simple: the model reads your data, it does not replace it. Use it to compress a hundred transcripts into a pattern, never to imagine a buyer you have never met.
Personas built from guesses, not deals?
Book a free 30-minute audit. We will pull your won deals, map the buying group hiding in your pipeline, and show you the three persona fixes we would make first.
Book an audit →Keep them alive
A persona is not a one-time deliverable. Markets move, buyers change jobs, the metric your economic buyer answers for shifts when their board changes strategy. The Gartner conflict number matters here too: when 74% of buying teams are arguing internally, the balance of power inside the committee is always moving. The persona that was right last year describes a room that has since reshuffled.
Set a quarterly review. Pull the last quarter's won and lost deals, check whether the roles and objections still match what your personas say, and edit. It takes an afternoon. Tie it to your win-loss process if you run one, because the same interviews feed both. A persona reviewed four times a year stays a working tool. One printed once becomes a poster.
The teams that get this right do not have prettier personas. They have shorter, uglier, more specific ones that name real objections and sit inside the CRM where the work happens. That is the whole difference. Stop describing a person who does not exist, start describing the five who decide your deals, and put the answer where your reps can reach it.
Frequently asked questions
What is the difference between a buyer persona and an ideal customer profile?
A buyer persona describes a person in the buying group: their role, pain, buying trigger, and decision criteria. An ideal customer profile describes a company: industry, size, revenue, tech stack, and growth stage. The ICP tells you which accounts to target, the persona tells you who to talk to inside them and what to say. You need both, and they should agree with each other.
How many buyer personas should a B2B company have?
One per distinct role in your buying group, not one per job title. For most B2B purchases that means four or five: champion, economic buyer, technical evaluator, end user, and sometimes a blocker. If you have built twelve personas, you have split by demographics instead of by decision-making role, and reps will ignore all of them.
How do I build a buyer persona with real data?
Start with 20 to 30 closed-won deals from your CRM. Interview five or six real buyers per role about what triggered the search, who else was involved, and what nearly stopped the deal. Mine your call recordings for the exact language buyers use and the objections they raise. Then write one page per role: trigger, the metric they answer for, first objection, and the line that moves them.
How often should buyer personas be updated?
Every quarter. Pull the last quarter of won and lost deals, check whether the roles and objections still match your personas, and edit. Only 44% of teams refresh personas annually, so even a quarterly review puts you ahead. Tie it to your win-loss review since the same interviews serve both.
Can I use AI to create buyer personas?
Use AI on your own data, not to invent buyers. A generic model will write a polished persona that is really just averaged internet content, which is the fiction you are trying to escape. Feed it your real transcripts and won-deal notes instead, and let it cluster objections and surface repeated phrases. The model should read your evidence, not replace it.
Build personas your reps will actually use
If your personas are a poster on the wall, they are costing you deals you cannot see, the ones that stall because nobody named the blocker or armed the champion. We help B2B teams build personas from their own won deals, map the buying group hiding in their pipeline, and wire it all into the CRM so it gets used on every deal. If that sounds like the fix you need, talk to us and we will start with a look at your last quarter of closed deals.