Why most personas are dead on arrival
The standard persona document looks like this: a stock photo of a smiling woman, a name like "Marketing Mary," her age (35), her job title (VP of Marketing), her favourite tools (HubSpot, Slack, Notion), and three bullet points about her pain points ("needs to scale content," "wants better attribution," "is busy"). Sound familiar?
That document does nothing. The next time the team writes a landing page, no one opens it. The next time the team picks a channel, no one references it. It's a slide, not a tool. The reason is simple: nothing on it tells you what changes Mary's mind. You can't write a headline that lands on her, because everything you know about her is true of fifty thousand other people.
A working persona has to do one job: when the team is about to make a decision (which channel, which message, which feature), the persona has to tilt that decision in a specific direction. If it doesn't, it's decoration.
The Jobs-to-be-Done shift
Clayton Christensen's Jobs-to-be-Done framing reframes the question. Instead of "who is this person?" you ask "what job is this person hiring our product to do?" The famous example is the milkshake — McDonald's discovered their morning milkshake buyers weren't kids being bribed; they were commuters who needed a one-handed, slow-to-finish, slightly-filling thing to keep them occupied during a 25-minute drive. The job was "make my commute less boring without spoiling lunch." Once you knew the job, every product decision changed: thicker shake, more chunky bits, smaller cup, sold near the drive-through window.
For B2B, the shape is the same but the jobs are bigger. A VP of Marketing isn't hiring your analytics tool because she likes dashboards. She's hiring it to win her next quarterly review, or to defend a budget she's about to lose, or to prove the new channel was worth the bet. Three different jobs, three different headlines, three different sales calls — even if Mary's title and tools are identical in all three.
The rule of thumb: if you can fit your persona's "job" into the sentence "I'm trying to ___, but ___, so I want to ___," you're close. If you can't, keep digging.
The ten conversations rule
You can't desk-research your way to a useful persona. The information you need lives in customer mouths and customer notebooks, and the only way to get it is to talk to ten of them.
Why ten? It's the smallest number where patterns become visible. After three you have anecdotes. After ten you start to hear the same phrase from three different people, and that phrase usually ends up in a headline.
Mix the ten:
- Three recent customers who said yes — what triggered them?
- Three current evaluators deciding right now — what's still in their way?
- Two churned customers who left — what did the product fail to do?
- Two who said no at the last stage — what did they pick instead?
Thirty-minute calls. Recorded with permission. The single best book on how to run them is Rob Fitzpatrick's The Mom Test — the core lesson is to ask about behaviour, not opinions ("walk me through what you did last Tuesday when…" beats "would you use a tool that…"). Behaviour is data; opinions are noise.
Anatomy of a persona that works
After ten conversations, you have raw material. Distil it into a one-page document with these sections — and skip the photo:
The job, in their words
One sentence in the customer's actual phrasing. Not "drive marketing efficiency." Something like "stop my CEO asking me every Monday what we got from LinkedIn this week." Specific, concrete, slightly ugly. That ugliness is the signal it's real.
The trigger
What event made them start looking? A new boss, a failed quarter, a competitor's launch, an outage, a board meeting. People don't search until something pushes them. If you know the trigger you can both target the moment and write copy that says "we know how you got here."
The current alternative
What are they doing right now instead? In B2B the answer is usually "a Google Sheet, three browser tabs, and complaining about it." Sometimes the alternative is a competitor; often the alternative is doing nothing. You're not just competing with rivals — you're competing with inertia.
The anxiety
What scares them about switching to your thing? Implementation pain, looking foolish in front of the team, lock-in, having to rebuild reports. Every anxiety is a sentence the landing page has to neutralise. If you don't know the anxieties, your CTA is shouting into a wall of "yeah but."
The decision unit
In B2B, who else has to say yes? The persona is rarely buying alone. Note the second and third stakeholders by role and the question each of them asks. The persona's headline is for them; the second-page detail is for the people the persona has to convince internally.
What they read, watch, ask
One short list of channels, but specific. Not "social media" — "the LinkedIn feed of these three competitors and the marketing weekly podcast." This drives the channel plan. If you can't be specific here, you haven't done the interviews.
A real example, sanitised
From a B2B engagement I ran last year — names changed, but the shape is real:
Persona — "the new VP defending a channel bet"
"I told the board we'd make LinkedIn work in six months. We're at month four and it's not. I need a number I can present at the next QBR that isn't a death sentence."
- Trigger: Q1 review three weeks away. Last year's CMO left. They were the one who championed paid social.
- Current alternative: Doubling down on the same campaigns, wishfully reading vanity metrics, and avoiding the analytics dashboard.
- Anxiety: "If I switch tactics now and it doesn't work, I look like I panicked. If I don't switch and it doesn't work, I look like I doubled down on a losing bet."
- Decision unit: Self + CFO (asks about CAC) + CEO (asks about pipeline contribution).
- Reads / watches: Hubspot's pipeline blog, two specific LinkedIn growth creators, the metric-stack newsletter.
Look at how much that document tells the team. The headline writes itself ("Walk into the QBR with a real number"). The hero CTA writes itself ("Book the diagnostic — see what's working before next week's review"). The channel plan writes itself (creator partnerships on LinkedIn, not generic paid). And every objection in the FAQ section maps to a real anxiety from the interviews.
That's a persona earning its keep.
How to use it once you have it
The persona document is the starting point, not the destination. Three uses that matter:
1 · Headline copy + hero section
Steal the customer's own phrasing. The closer your headline is to what they say in interviews, the more it lands. "Stop your CEO asking what you got from LinkedIn last week" beats "Drive marketing ROI with measurable outcomes" by an order of magnitude — even though the second one sounds more professional.
2 · Channel selection
Channels follow the audience, not the trend. If your persona reads three specific LinkedIn creators and a niche newsletter, that's where you show up. TikTok is irrelevant to most B2B personas in 2026, no matter how loud the agency pitch is.
3 · Product/feature roadmap
The anxieties from interviews map directly to feature priority. If half your interviews surface "I'm scared the implementation will eat a quarter," ship a self-onboarding flow before you ship the next dashboard upgrade.
Reread the persona quarterly. Markets shift, triggers change, alternatives evolve. A persona that hasn't been updated in a year is a museum exhibit, not a tool. The /now page on this site is built around exactly this principle — keep the live state visible, refresh on a known cadence.
Mistakes I see most often
- Building five personas when you have one channel and one product. If you can't run different campaigns for them, they're not separate personas — they're variants of the same one.
- Inventing personas instead of interviewing for them. "We think our customers are…" is the most expensive sentence in marketing. The real customers are an email away.
- Confusing demographics for psychographics. "VP, age 38, in fintech" doesn't tell you what to write. "Just took the role and is presenting numbers in three weeks" does.
- Writing the persona in your voice, not theirs. Steal their phrases verbatim. Marketers polish the language, and the polish removes the value.
- Treating the persona as one-and-done. Re-interview every six months. The anxiety set shifts faster than you'd think.
- Skipping the churned customers. The churned and the "no" customers tell you more about why you lose than any won deal ever will.
Practical checklist
If you have an hour, do step one. If you have a fortnight, do all of it.
- ☐ List ten people to call: 3 recent yes, 3 currently evaluating, 2 churned, 2 said no.
- ☐ Email them with a 30-min ask, recorded with permission, no sales agenda.
- ☐ Use Rob Fitzpatrick's Mom Test question pattern: behaviour over opinion.
- ☐ Transcribe verbatim. The phrasings are the gold.
- ☐ For each: what triggered the search, what they tried first, what scared them about switching, who else had to say yes.
- ☐ Cluster across all ten. The patterns that show up 4+ times are your persona.
- ☐ Write the one-page persona using the structure in this article. Skip the photo.
- ☐ Pull three customer phrases verbatim and turn them into headlines. A/B test.
- ☐ Re-read every quarter. Re-interview every two quarters.
- ☐ When the team is about to make a decision (channel, copy, feature), open the persona doc before the meeting starts.
Done well, a persona is the cheapest way to make every other marketing decision in the company more right than wrong. Done badly, it's a slide. The difference is ten conversations.