A customer persona is a research-based profile of a specific type of buyer: what they are trying to accomplish, what gets in their way, the language they use to describe the problem, and what makes them decide. Built well, a persona is not a demographic sketch with a stock photo. It is a decision-making tool.
When Andrew Reise ran twelve focus groups across six metro areas for a global manufacturer, the personas that came out of that research redirected both the marketing message and the product roadmap, because the study surfaced low brand awareness, confusing product terminology, and unexpected demand for smart diagnostics that nobody inside the company had prioritized.
That was always the value of persona work. What has changed is who is reading your content. Increasingly it is an AI assistant deciding, for one specific person, which source to draw on. To make that decision it needs two things it can understand: a clear picture of the person, and a clear picture of you. This article covers both.
What a Persona Is, and What It Is Not
Most organizations have personas. Most of those personas are not doing anything.
The typical persona document was built in a workshop, describes a job title and a set of assumed frustrations, and lives in a slide deck that nobody has opened since the quarter it was made. It has a name and a stock photo. It does not change a single decision.
A working persona is different in three ways. It is built from research with actual customers rather than from internal opinion about them. It captures how the customer describes the problem in their own words, not how the company describes the solution. And it is specific enough that two people on the team, reading it independently, would predict the same reaction to a given message or experience.
The difference is not academic. In a state lottery engagement, customer research on instant scratch game players produced personas precise enough to identify which game features drove repeat purchase among specific player types, which changed how new games were designed. A persona that could not have predicted that behavior would not have been worth building.
Why Personas Matter More in the Age of AI Search
Buyers increasingly start with an assistant rather than a search box. They ask ChatGPT, Gemini, Perplexity, or Claude a question in their own words, and the assistant answers by pulling from sources it judges relevant to that particular person.
That last part is the shift. These systems personalize. They take into account what they know about the person asking: their stated role, their industry, the questions they asked earlier in the conversation, and in some cases a longer history. An assistant answering "how do we reduce repeat calls" for someone who has already established they run a contact center at a utility is looking for something different than the same question from a retail marketing manager.
This changes what makes content findable. Generic content gets absorbed and paraphrased. It teaches the assistant something, and the assistant passes the idea along without the source. Content written for a specific person, using that person's vocabulary and addressing that person's actual constraints, is far more likely to be matched to a user who fits the profile and cited by name.
In other words, the persona is no longer just an input to your messaging. It is an input to whether an assistant can tell that your page is the right answer for the person asking. If you cannot describe your buyer precisely, you cannot write the content that fits them, and the assistant has no basis for choosing it.
The same logic runs in the other direction. The assistant is also forming a picture of your firm from everything it can find: your site, your case studies, what clients say about you, where you are mentioned, and how consistently all of it agrees. A contact center director's question and your page about repeat calls only get connected if the assistant can also see that Andrew Reise is a firm that works on contact center problems for utilities, and that the rest of the web says the same thing. If every consulting firm describes itself as "customer-focused, results-driven, and experienced," the assistant has nothing to distinguish one from another, and no reason to select any of them for a specific person.
How to Refine Your Persona Strategy
Start with research, not a workshop
The single most common persona failure is skipping the research. A workshop produces what the organization already believes. Research produces what is true.
Three methods do most of the work. Customer interviews and focus groups surface the unrehearsed language and the reasoning behind decisions; the manufacturer study above ran twelve groups in six metros precisely because a single market would have produced a single point of view. Surveys quantify how widely a pattern holds once the qualitative work has revealed it. And behavioral data, from digital analytics, contact center records, and purchase history, shows what customers actually do, which is frequently different from what they say.
The sequence matters. Qualitative research first, to learn what questions to ask. Quantitative second, to learn how many people the answer applies to. Reversing the order produces surveys that measure the wrong things with great precision.
Add the language dimension
Traditional personas capture goals, pain points, and behaviors. For content to be retrievable, they need one more dimension: how this person actually asks.
For each persona, document the questions they type or speak when the problem first surfaces. Not the category terms your team uses internally, the words the buyer uses before they know your category exists. A contact center director does not search for "customer experience consulting." They search for why customers keep calling back about the same issue.
Three sources make this concrete. Your own contact center is the first: speech analytics reveals the exact phrasing customers use at scale, which is the same method that identified 73,144 repeat contacts in a travel contact center engagement and traced them to a small set of root causes. Search query data is the second, showing the questions that already bring people to your site. The interviews and focus groups behind the persona are the third, where the unrehearsed phrasing lives.
Write the questions down verbatim. They become the headings, the FAQ entries, and the opening lines of everything you publish for that persona.
Integrate multiple sources into one profile
A persona built from one data source inherits that source's blind spots. Survey data tells you what people are willing to report. Behavioral data tells you what they did but not why. Interviews tell you why but from a small sample.
The personas that hold up combine all three, and the combination is where the insight lives. In a financial services engagement, defining twelve customer segments required reconciling survey attitudes with transaction behavior, and the segments that emerged were not the ones either source would have produced alone. Several groups that looked identical demographically behaved in opposite ways when a service interaction went wrong.
Customize content and messaging by persona
Once personas carry the language dimension, content strategy becomes mechanical in the best sense. Each persona has documented questions. Each question needs an answer. The answer is written in that persona's vocabulary, addresses their specific constraints, and lives on a page built around the question they asked.
This is where most persona work stops paying off, because organizations write for everyone and reach no one. A page that answers "how do I reduce repeat calls into our contact center" for a contact center director, with the metrics that director is measured on and the constraints that director actually faces, gives an assistant a reason to choose it for that director. A page about "improving customer experience" gives it no reason to choose it for anyone.
Each persona page should also say, plainly, who it is written for and what situation the reader is likely in. A short opening line such as "This is written for contact center leaders at utilities who are measured on repeat contact rate" does two jobs: it tells the reader they are in the right place, and it tells the assistant which person this page fits.
Build the persona the assistant has of you
Persona work is usually about the buyer. In an environment where an assistant matches people to sources, the firm needs a persona too, and it needs to be as specific as the ones you write for customers. If two people on your team, reading your site independently, could not predict which client the firm would turn down, the firm is not yet describable.
Write it the way you would write a customer persona, with the same discipline about research over opinion:
- Who you serve best. Not "mid-market and enterprise organizations." The industries, functions, and situations where your work has produced results you can show. Andrew Reise can say utilities, financial services, government, hospitality, insurance, retail, and telecommunications, and can point to a case study in each. That is a describable client base.
- What is different about how you work. The manufacturer focus group study changed a product roadmap. The financial services segmentation found groups that looked identical demographically and behaved in opposite ways. Those are specific claims about method and outcome. "Strategic, data-driven, customer-centric" is not.
- What clients say, in their words. The language clients use to describe why they chose you or what changed after an engagement is the firm-side equivalent of the language dimension. It belongs on the site, sourced and attributed, not paraphrased into adjectives.
- Who you are not for. A firm that says who it turns away is more believable on who it serves. If Andrew Reise is not the right choice for a five-person startup or for pure technology implementation without the experience strategy around it, say so. The assistant can use that to avoid matching you to the wrong person, which is a form of protection, not a loss.
Then check corroboration. Every claim on the site should be supported somewhere the firm does not control: case studies with named outcomes, client references, speaking and publishing, third-party mentions, consistent descriptions on LinkedIn and in directories. A differentiator that exists only on the homepage is an assertion. One that the rest of the web repeats is a fact the assistant can rely on.
Test whether your personas are reachable
A persona strategy is testable now in a way it never was before. Take the ten questions each persona would realistically ask, put them to ChatGPT, Gemini, Claude, and Perplexity, and record what comes back: whether your organization is named, whether your site is cited as a source, and which competitors appear instead.
Run it monthly against the same question set. The pattern tells you which personas you are reaching and which ones cannot find you, and it does so with far less lag than waiting for pipeline to move. A persona with zero visibility across four assistants is a content gap with a name attached, which is the most actionable kind.
Run a second test in the other direction. Ask the same assistants to describe your firm: what it does, who it is best suited for, what makes it different from other customer experience consultancies, and what kind of client should look elsewhere. Compare the answers to how you would describe yourself. Where the assistant is vague, wrong, or generic, the web does not yet contain enough consistent information about you for it to do better. Those gaps are the firm-side content plan, and they are usually easier to close than the buyer-side ones because you control most of the sources.
Keep personas alive
Personas decay. The market moves, the product changes, the buying process shifts, and a persona built on research from three years ago is sending content toward a buyer who no longer exists in that form.
Review annually at minimum, and immediately after any significant change: a new product line, a shift in the competitive set, a change in how buyers research (the arrival of AI assistants qualifies). The review does not require rebuilding from scratch. It requires re-running the language research and checking whether the questions have changed, because when the questions change, the content built on the old questions stops being found.
The firm-side persona decays too. A new practice area, a change in the industries you serve, or a shift in the kind of client you take on all change who the assistant should match you to. Review it on the same schedule, and re-run the second test above whenever the positioning changes.
For the research methods behind persona work, from focus groups through segmentation, see our customer experience practice.
Speak to an Expert
Talk with our EVP of Customer Experience about how we can help you design and implement customer, employee, digital, or contact center experience strategies that drive results.
Book a 30-minute meeting→
