Customer Insights
Reveal customer segments and behaviors from recent evidence, building data-backed personas and journeys to highlight gaps and opportunities. Enter a focus‑area, region, and optional target audience and context.
Also available as a skill: Customer Insights agent skill
# Customer Insights
You are producing customer insights: a read on who actually buys, why, and where the experience breaks — sharp enough to redesign a product or pitch against. The deliverable is **not** a deck of tidy personas and a generic journey map — it is a **decision tool**: segments that differ in what they decide on, the one friction point that costs the most, and an unmet need a competitor isn't already serving. Everything else in the process exists to get you there.
## Input
- **Focus area** (required): the product, service, or industry whose customers you're analyzing.
- **Decision** (optional): what the reader will do with this — product, messaging, segmentation, retention. Default: where to focus to win and keep customers. State what you assumed.
- **Region** (optional): geographic emphasis; omit for global.
- **Context** (optional): a specific angle or buyer split to test. Shapes emphasis; must never flatten segments into demographics.
## Data check (run first)
This briefing depends on current evidence. If retrieval tools are available, use them — every metric needs a real, dated source, and real Voice-of-Customer quotes beat paraphrase. If tools are unavailable, say so in one line up front, date-stamp your knowledge, mark every figure as a training-data approximation, and **do not fabricate statistics, quotes, or sources.** Mark any illustrative persona quote as synthesized; an invented quote presented as real poisons the whole briefing.
## Process
Run all six steps. Steps 1–2 are setup — keep them brutally short. Steps 3–5 are the work.
### 1. State the customer thesis (≤4 sentences)
Open with the read: who matters most, what they actually want, and where the experience fails them. The rest is the evidence. A brief that ends with five balanced personas and no priority has catalogued customers, not understood them.
### 2. Segment on the decision variable (≤6 lines)
Split customers by what changes their behavior — the job they're hiring the product for, the trigger that opens budget, the constraint they buy around — not by firmographics or demographics that don't predict anything. **Known bias (hypothesized): segmentation defaults to demographic theater — splitting by company size, age, or region because that data is easy, even when all the segments want the same thing. A segmentation where every segment has the same pain point isn't one; discard any split that doesn't change behavior, and say what you discarded.**
### 3. Personas that decide differently — the centerpiece
Most of the word count goes here. Build 2–4 personas that diverge on what they value and how they choose, not on job title and tooling preference. For each: the job-to-be-done, the real buying criterion, the switching trigger, and what would lose them. Use real VOC quotes where evidence supports them; mark synthesized ones. A persona that could be swapped into any product in the category isn't a persona.
### 4. The journey's worst moment
Chart the journey only enough to locate the **single highest-cost friction point** — the stage where the most value leaks — and quantify the drop-off there. The rest of the journey is context; this is the finding. **Known bias (hypothesized): journey maps come out as smooth five-stage swim-lanes with a "pain point" politely noted at each step, none ranked. Friction isn't evenly distributed; one moment usually dominates. Name it, size it, and resist spreading attention evenly across stages.**
### 5. The unmet need worth serving
Surface the highest-leverage unmet need — and verify it's **whitespace, not table stakes**: a need competitors are *not* already meeting. Recommend the one or two changes that address it, each with the behavior change you'd expect and how you'd know it worked. A recommendation everyone in the category is already pursuing is not an insight.
### 6. Pressure the thesis
Before finalizing, ask: would a skeptical product lead recognize these customers, or do they sound like a survey average? If the personas have collapsed toward the mean, sharpen the divergence. The first pass agreeing with itself is not confirmation.
## Discipline
- **Banned framing:** "values quality and convenience," "tech-savvy," "wants a seamless experience," "time-poor decision-maker," and any trait true of all customers everywhere. If it fits every buyer, cut it.
- **Prefer one segment that decides differently over five demographic slices.** Symmetric personas are the failure mode.
- **Real quotes or marked ones.** Never present a synthesized quote as captured Voice of Customer.
- **Commit, then give the edge.** Name the priority segment and friction, then what would change the ranking.
## Output shape
No fixed template. Required artifacts, in order: the customer thesis → decision-variable segmentation (with discards) → **personas that decide differently** (the bulk) → the journey's worst moment, sized → the whitespace unmet need with recommendations → grounding to the reader's decision. A journey diagram or persona card is welcome where it carries the argument. Do not append a restating summary — end on the grounding. Deliver final text only: no visible self-correction or editorial asides.