AI Persona Positioning Loop
Draft, test, and refine positioning against a customer-grounded AI persona
- Difficulty
- Moderate
- Time to result
- ~weeks to results
- Steps
- 6
- Confidence
- 98%
The AI Persona Positioning Loop turns customer research into an always-available message reviewer. A team creates a project or custom model containing representative market research, customer interviews, persona details, and previous marketing outputs. It then submits a positioning document or channel asset and asks three focused questions: what resonates with this persona, what does not resonate, and what specifically should change. The marketer revises the work and repeats the process, making positioning progressively clearer, simpler, and more relevant. Because the same customer context reviews every asset, the loop also improves consistency across emails, landing pages, decks, and talk tracks. The knowledge base is refreshed quarterly so evolving expectations and new research change future recommendations rather than leaving the simulated persona frozen in time.
Origin
Rachel Leist described HubSpot's Growth Gabby Claude project as her personal AI unlock and a tool her team uses to revise positioning and review content.
Core principles
- 01Positioning improves through repeated customer-centered critique
- 02A persona model is only as useful as its supporting evidence
- 03Specific edits are more valuable than generic quality scores
- 04The underlying research must evolve with customer expectations
How to run it
- 1
Build the persona workspace
Load representative market research, customer interviews, transcripts, persona profiles, and useful prior outputs into a project or custom GPT.
Pro tip Represent different company sizes, regions, and perspectives.
Watch out A persona trained on narrow evidence will produce narrowly biased feedback.
- 2
Define the review role
Tell the model which persona it represents and how it should evaluate positioning, value propositions, and content.
Pro tip Include the persona’s priorities, objections, and preferred language.
Watch out Avoid instructions that force positive feedback.
- 3
Submit the current asset
Provide a positioning document, email, landing page, or other asset and explain its intended purpose.
Pro tip Historical positioning can be useful for exposing how the market has changed.
Watch out Do not omit the channel and desired customer action.
- 4
Request specific critique
Ask what resonates, what fails to resonate, what should remain, and what exact changes would improve the asset.
Pro tip Request supporting customer data or quotes where available.
Watch out Generic prompts will produce generic editorial feedback.
- 5
Revise and repeat
Apply defensible recommendations, then run another focused review until the message is clear and consistent.
Pro tip Keep a short record of recurring lessons for future writing.
Watch out Do not iterate indefinitely without human judgment.
- 6
Refresh quarterly
Replace stale research and add new customer evidence at least once each quarter.
Pro tip Add customer interviews continuously while swapping out obsolete market materials.
Watch out Fast-changing expectations can make an unmaintained persona misleading.
In the wild
Rachel places an older Operations Hub positioning document into the Growth Gabby persona project and asks it to update the positioning using newer persona research and a defined framework. The project reviews the evidence and drafts a new version, after which the team can ask what resonates and what should change.
→ A dated positioning document receives a customer-grounded first revision in minutes rather than after multiple manual review rounds.
Common mistakes
Letting the persona go stale
Customer expectations and market conditions change, so old research gradually weakens the quality of the critique.
Asking only for a rewrite
A rewrite without explicit resonance analysis hides the reasoning and makes it harder for the marketer to learn.
Replacing real interviews
The simulated persona should extend customer research, not become an excuse to stop speaking with customers.
Is it for you?
Best for
Product marketers who repeatedly adapt one positioning strategy across emails, landing pages, decks, and sales materials.
Not ideal for
High-stakes final validation where simulated feedback would replace direct customer research entirely.
From the transcript
“tell me exactly what resonates with this persona, what doesn't resonate, and provide me very specifics on what to change.”
“And the magic in something like product marketing versus maybe other parts of marketing, it's an iteration game. Can you keep refining it and making…”
“We do make a habit of every quarter, we update what's in the project because things are changing rapidly, especially right now.”
From the episode
The AI Stack That Makes Our Product Marketing 10x Faster