Buyer personas and campaign planning

How do you build buyer personas and campaign plans that still work when AI answers first?

The shortlist is set before you are invited. In 6sense's 2025 B2B Buyer Experience Report, a study of about 4,000 buyers, 94 percent of buying groups had already ranked their preferred vendors before making first contact, and 77 percent bought from that preliminary favorite. The winning vendor was on the Day One shortlist 95 percent of the time.

That changes what a persona is for. The persona you build today is not only a guide for your team. It is the brief an AI engine uses when it decides whether your brand belongs in the answer.

Most persona templates still in circulation were built for a journey your brand controlled: a name, an age, a stock photo, a job title, and a set of pain points. The room has gotten bigger, the research now starts inside a generative engine, and the questions buyers type are longer and more specific than any keyword you tracked.

This guide covers what a buyer persona needs to contain in 2026, how to build one in five layers, how to turn those layers into a campaign plan, and how to judge whether it worked.

What you will take away

• The difference between an ideal customer profile, a buyer persona, and a buying group map
• The CCG Persona Stack, a five layer build that ends with a tracked AI prompt set
• A buying jobs map that tells you what proof each person needs, and when
• The Loop Campaign Planner, with sequencing rules that keep spend behind proven content
• The budget math behind brand building and lead generation, with the formula shown
• Honest evaluation windows, so you judge campaigns on the right clock

A note on the numbers before you read further. Every statistic on this page carries a named source and a year. Several of the most repeated persona statistics on the internet trace back to a single 2015 survey or to case studies with no disclosed method, so they are not here. Conspicuouz Creative Group cites what can be traced.

The build

How do you build a buyer persona step by step?

Build the persona in five layers: the ideal customer profile, the buying group map, role personas, buying jobs, and the prompt set. Most published templates stop at layer three. The last two layers connect a persona to the work a buyer is trying to finish and to the exact questions they put into an AI engine.

The CCG Persona Stack

Layer 1: Write the ICP in one page

1. 1Pull your closed won accounts from the last 24 months and sort by retention and margin, not by deal size alone.
2. 2Find the shared characteristics of the top third: revenue band, vertical, geography, team structure, existing tools.
3. 3Write the disqualifiers next to the fit criteria. Name the accounts you should decline.
4. 4Keep it to one page. If sales cannot apply it on a first call, it is too long.

Layer 2: Map the room

1. 1List every person who touched your last five won deals and your last five lost deals.
2. 2Assign a role to each: champion, economic buyer, technical validator, hidden buyer, external influencer.
3. 3Mark who never appeared on a call but shaped the outcome. Those are the hidden buyers, and they are usually in finance, legal, procurement, compliance, or operations.
4. 4Note where the group stalled. Forrester (2026) found 94 percent of buyers in groups of six or more report clear benefits from the group, including broader perspective, shared validation, and budget security. The group is not the enemy. Ignoring most of it is.

Layer 3: Interview real people

Six to ten conversations per core role is enough to find the pattern. Record them, and keep the exact phrasing.

Questions worth asking, organized by buying job:

• Problem identification: What was happening in the business the week you decided this needed solving?
• Solution exploration: Where did you look first, and what did you type?
• Requirements building: What did the shortlist have to do before it made the list?
• Supplier selection: What nearly disqualified us?
• Validation: Who did you check with before you were comfortable?
• Consensus creation: Who did you have to convince internally, and what did you send them?

Layer 4: Lay the buying jobs over each role

That last row has support behind it. Edelman and LinkedIn (2025) found 71 percent of hidden decision makers consider thought leadership a more trustworthy basis for evaluating a vendor than product sheets or marketing materials, 86 percent prefer content that challenges their assumptions, and 79 percent are more likely to champion vendors with consistently strong thought leadership. For a challenger brand, one finding in that report matters most: 53 percent of decision makers say strong thought leadership makes brand recognition matter less.

Layer 5: Build the prompt set

Layer five is where the persona stops being a document and starts being a test. It records the exact questions each role types into a generative engine, tested across all five major engines and re run on a schedule. The next section covers how to build it.

What to remove from the old template

Stock photos. A fictional first name as the headline. Hobbies unrelated to buying. Long invented backstories. Research on AI generated personas, covered below, found that added fictional detail makes predictions worse, and the same logic applies to the ones your team writes by hand.

What to add

• Trusted validators: the people, publications, and communities they check answers against
• Buying job status: what this person is trying to finish right now
• Exact prompts: the real phrasing, not a tidied up version
• Disqualifiers: who in this role you should not pursue

Synthetic personas

Can AI generated or synthetic personas replace customer research?

AI generated personas sharpen research. They do not replace it. Maier and colleagues (2025) reached about 90 percent of human test retest reliability simulating consumer survey responses using a validated semantic similarity method across 57 surveys and more than 9,300 human responses, but that result holds for structured tasks with a defined method, not for open ended discovery.

The evidence on where synthetic personas break down is unusually clear.

• Added fiction makes them worse. Work by Li, Chen, Namkoong and Peng of Columbia, presented at NeurIPS 2025, found richly detailed LLM generated personas performed worse than minimal, census style personas at predicting the 2024 US election.
• Persona descriptions added nothing to forecasts. Iadisernia and Camassa of Banca d'Italia, at ICAIF 2025, ran an experiment covering 118,400 forecasts and found persona descriptions produced no measurable forecasting advantage.
• Simulated respondents are too agreeable. A Kromatic review of peer reviewed studies (March 2026) found LLM simulated respondents skew more positive, more progressive, and less varied than real people.
• Most studies do not define the job. A 2025 review of 63 peer reviewed persona studies found task and target population are frequently underspecified.
• They are weakest where B2B is hardest. PyMC Labs' practitioner guide (February 2026, vendor source, directional) places synthetic consumers strongest in structured tasks such as ranking, pricing, and sentiment, and weakest in emotional nuance, culture, and group dynamics.

Group dynamics is exactly what a B2B buying group is made of. A tool that is weakest at consensus behavior is the wrong tool for modeling a room of 13 internal stakeholders and 9 external influencers.

Campaign planning

How do you turn personas into a campaign plan?

Campaign planning turns persona evidence into sequence. Express sets the point of view, Tailor matches the message to the buying job, Amplify places it where each persona discovers and validates, and Evolve feeds what you learned into the next cycle. The persona is the input at every stage, not a document you consult once at kickoff.

HubSpot introduced Loop Marketing, built on Express, Tailor, Amplify, and Evolve, at INBOUND 2025 as a replacement for the funnel and the flywheel. CCG runs client engagements on that Loop inside a 90 day structure, with GEO infrastructure built first.

The Loop Campaign Planner

Three sequencing rules worth stating plainly

1. 1No content before Express is complete and approved in writing. Content produced before the point of view is settled becomes rework.
2. 2Paid budget only goes behind content that proved itself organically first. Paying to distribute an unproven message buys you a faster, more expensive no.
3. 3Infrastructure before content volume. Entity clarity, technical foundation, and consistent facts across every property come first. The gap closes with infrastructure, not more content alone.

See how the Loop runs inside a 90 day engagement, and what the 90 day GEO retainer covers.

Start here

Build the persona your buyers are actually using

Personas built for a journey you controlled will not hold up in a market where the shortlist forms before the first call and the first answer comes from an engine. The fix is orderly: map the whole room, name the job each person is finishing, capture the language they use, and build the infrastructure that lets an engine cite you when they ask.

Conspicuouz Creative Group builds that infrastructure for growing brands, then runs the Loop on top of it.

Get your free AI Visibility snapshot, delivered by email within one business day. It shows where your brand appears across ChatGPT, Claude, Gemini, Perplexity, and Google AI Overviews for the questions your buyers are already asking, and where your competitors are being cited instead.

Request your AI Visibility snapshot