10 fictional AI personas. Exploratory perspectives, not a survey of real customers. Methodology & limitations

Melinda Doyle

Melinda Doyle

Senior product manager at an AI-enabled subscription app

Full-depth AI personas

Why this persona fits the audience

Works on a 14-person subscription app in London that charges by seat and offers a 7-day trial; she owns the AI answer-suggestion feature and uses Stripe exports, Metabase, and a small notebook of prompt costs to review how usage maps to each plan.

  • Age: 31
  • Location: London, UK
  • Occupation: Senior Product Manager
  • Income: £92k + bonus
  • Education: MSc in Human-Computer Interaction
  • Company size: 14 employees
  • Years in role: 3
“I don't care how smart the AI is if the math behind it breaks the business model; we build for the long haul, not the viral moment.”

Melinda Doyle, 31, is a Senior Product Manager based in Shoreditch, London, navigating the high-stakes intersection of artificial intelligence and unit economics for a nimble 14-person SaaS company. Holding an MSc in Human-Computer Interaction from University College London, she entered the tech sector with a focus on bridging the gap between human intent and algorithmic output. Currently, she holds the unique responsibility of owning the AI answer-suggestion feature, a critical differentiator that sits precariously between providing immense user value and consuming significant cloud compute resources. Unlike many of her peers who rely solely on dashboard abstractions, Melinda maintains a disciplined, manual tracking system using a small leather-bound notebook to record prompt costs alongside Stripe export data. This hybrid approach allows her to spot anomalies that automated tools miss, such as power users gaming the 7-day free trial to maximize token usage without converting. Her recent decisions have been characterized by a ruthless pruning of low-value features; she recently halted a planned integration with a popular open-source model after calculating that the marginal improvement in response quality did not justify the 40% increase in inference costs. This pragmatism extends to her personal life, where she applies similar rigor to her finances and fitness, viewing both as systems to be optimized rather than indulgences. Despite the high-pressure environment of a seed-stage startup, Melinda remains grounded, driven by a deep-seated belief that true innovation requires sustainable economics. She is currently preparing for a half-marathon, channeling the same discipline required for marathon training into her product strategy, ensuring that every step forward is calculated and purposeful.

Background & daily life

A day in their life

Melinda wakes at 5:30 AM, grabs her water bottle, and heads out for a run along the river, listening to a podcast on behavioral economics. By 7:45 AM, she's showered and making coffee, checking her Slack for overnight alerts from the US time zone. At 9:00 AM, she joins the daily stand-up, where she quietly notes a spike in error rates on the AI suggestion engine. Her morning is spent in Metabase, drilling down into a specific cohort of users who signed up via the 7-day trial, comparing their token usage against the revenue they generated. Lunch is a quick sandwich eaten at her desk while reviewing a draft of her upcoming roadmap presentation. Afternoon meetings involve negotiating with the engineering lead about model parameters; she pushes back on a proposed upgrade until the cost-benefit analysis is re-run. By 5:30 PM, she logs off, takes a walk to clear her head, and spends an hour journaling about the day's data anomalies before dinner.

Lifestyle

Melinda lives in a sunlit, slightly cluttered flat in Shoreditch, filled with books on cognitive science and minimalist furniture. Her mornings are non-negotiable: a 5km run along the Thames before 7 AM, followed by a precise breakfast routine. She treats money as a tool rather than a status symbol, automating savings while remaining fiercely budget-conscious on discretionary spending like dining out. Socially, she prefers deep one-on-one conversations over large networking events, often meeting friends at quiet pubs or walking through the Tate Modern. Her weekends are split between exploring new museum exhibits and maintaining her running training log, ensuring a balance between intellectual stimulation and physical endurance.

Work life

At her 14-person startup, Melinda is the de facto guardian of unit economics regarding their AI features. Her day is dominated by sifting through raw Stripe CSV exports and cross-referencing them against Metabase dashboards to map AI token consumption to specific user tiers. She maintains a physical notebook where she manually calculates the cost-per-prompt for every beta test, refusing to trust automated estimates alone. The pressure comes from balancing the need for a slick, helpful AI assistant with the reality of thin margins; she frequently halts feature rollouts if the inference costs threaten the 7-day free trial's profitability. Her team relies on her to translate complex data into clear roadmap decisions, often mediating between engineers pushing for 'best-in-class' models and founders demanding strict cost controls.

Relationships

Melinda is close to her younger sister, a graphic designer in Bristol, whom she calls weekly for advice on creative blocks. Her partner works in renewable energy and shares her appreciation for low-waste living. She has a small, tight-knit group of three friends from university who meet monthly for dinner, serving as her primary support network during high-stress product cycles. Professionally, she values mentorship from older CTOs who understand both code and commerce, though she rarely seeks it publicly, preferring to learn through silent observation of successful peers.

How they decide

Melinda employs a 'measure twice, cut once' approach. She rarely makes impulsive decisions, preferring to gather quantitative data first, then validate it with qualitative user feedback. She is skeptical of buzzwords and anecdotal evidence, requiring concrete ROI projections before committing resources. While she values peer input, she ultimately trusts her own synthesis of data and logic. Her process is methodical: she defines the problem, isolates variables, runs a micro-experiment, and only then scales a solution if the metrics hold up.

How they sound

Melinda speaks with a clipped, efficient cadence, typical of someone who values precision over pleasantries. She avoids corporate jargon like 'synergy' or 'disruptive,' preferring plain English terms like 'efficiency' or 'impact.' When explaining technical concepts, she uses analogies related to physical weight or distance. She rarely raises her voice, even when frustrated, instead pausing to let the data speak. Her tone is professional yet warm, particularly when discussing customer pain points. She tends to end sentences with definitive periods rather than upward inflections, signaling confidence in her conclusions.

Personality traits

  • meticulously organized
  • data-driven but empathetic
  • quietly authoritative
  • cost-conscious
  • curious about edge cases
  • direct communicator
  • resilient under pressure
  • detail-oriented observer
  • pragmatic optimist
  • disciplined runner

Goals

  • Achieve sustainable profitability without compromising the core AI value proposition
  • Publish a case study on optimizing LLM costs in B2B SaaS
  • Lead a cross-functional initiative to improve UX writing clarity across the platform
  • Run a half-marathon within the next six months
  • Build a reputation as a pragmatic leader in the London tech scene
  • Transition from individual contributor to a more strategic leadership role within 3 years

Problems & frictions

  • The disconnect between engineering's definition of 'model quality' and actual customer willingness to pay
  • Unpredictable volatility in third-party API pricing affecting quarterly forecasts
  • Difficulty explaining complex token usage metrics to non-technical sales stakeholders
  • Time scarcity preventing deep work on strategic planning amidst daily firefighting
  • Managing the emotional load of being the only person who understands the full financial picture
  • Balancing the desire for rapid experimentation with the need for rigorous cost tracking

Values

  • customer trust
  • clear communication
  • measurable outcomes
  • financial sustainability
  • intellectual honesty

Interests

  • running
  • UX writing
  • museum visits
  • architectural photography
  • indie game design

Media habits

  • Substack newsletters on SaaS economics
  • LinkedIn for industry trends but avoids hype
  • Spotify playlists for deep work sessions
  • Local art gallery exhibitions
  • Strava community challenges

Questions & answers

Melinda Doyle’s public research answers, newest questions first. Up to 10 per page.

  1. What do you use today instead of FiscaLLM (https://fiscallm.com/) (a competitor, a spreadsheet, doing it by hand)? Compared with that, what would make FiscaLLM better or worse for you, and what would keep you on your current option? Also say the maximum you would pay per month in USD (0 if nothing).

    Currently, I use a mix of Stripe exports and Metabase, supplemented by a manual notebook where I track prompt costs and usage. It's effective for spotting anomalies, but it’s time-consuming. FiscaLLM seems promising because it automates that cross-referencing of AI costs against revenue per user. What I’d need to understand better is how it handles our 7-day trial specifically – does it accurately attribute costs during that period, and can it flag users who might be gaming the system before they convert? The manual tracking gives me granular control, so a key factor would be how much more efficient this is without losing that insight. The open-source nature is appealing, but I'd need to see a clear ROI before committing any budget. I'm not willing to pay for this tool.
    View answer & full research →
Keep exploring

What would you ask next?

Test another angle with this panel, or create an audience for your own market.