How Customers Talk, Search, and Buy

Three behaviours complete the discovery picture: the language customers use for their problem, the channels they trust when they go looking for solutions, and the process by which they actually decide to pay. Get these from evidence, not assumption—each one shapes surfaces you will ship.

What it is

Voice of the customer (VoC) is the customer’s own vocabulary for the problem, the solution category, and the outcome—captured verbatim from interviews, support tickets, sales calls, reviews, and community posts. It is data, not decoration: the exact phrases people use are how they recognise that a product is for them.
Solution-seeking behaviour is where and how people look when a need state activates: search queries (in their words), communities and peers they trust, review sites, app stores, and—increasingly—questions put to artificial intelligence (AI) assistants. These are the “watering holes” where your ICP already gathers.
The buying journey is the path from problem to payment: who is involved, what evidence they need, what they compare, what could get you disqualified, and what makes the risk feel survivable. In consumer products this may be minutes long and solo; in business products it is a buying committee—champion, economic buyer, end users, security review—each needing different reassurance.

Why it works

Each behaviour maps to surfaces with a feeling on the line:

Talk → recognition. Copy that mirrors the customer’s language produces the small, decisive feeling of being understood (“finally, someone gets it”); copy in your internal jargon produces the translation tax that Friction catalogues as cognitive cost. You cannot mirror language you never collected.
Search → arrival state. The channel someone arrives from encodes their awareness stage: a community recommendation arrives warmer than a cold search; a comparison-site click arrives evaluating. Designing one landing experience for all arrivals wastes what the channel already told you.
Buy → calibrated trust under risk. Buying is the moment doubt peaks: money, commitment, and the fear of choosing wrong converge. A checkout or upgrade surface designed without knowing the buyer’s real evaluation process—what they need to verify, whom they must convince, what burned them last time—will supply generic persuasion where specific reassurance was needed (The Paywall, Trust Building, Jobs-to-be-Done).

Going deeper

Mine language before writing any. Harvest verbatims from interviews, tickets, reviews of competitors, and community threads; cluster them by need state; note the words customers use and the words they never use. Your category jargon (“workflow orchestration”) versus their words (“stop babysitting the hand-offs”) is a measurable gap—close it in their favour (JTBD Copywriting, Product Voice).
Follow the actual search, not the imagined one. In interviews, reconstruct the last real search: what did they type, whom did they ask, which answer did they trust, why? Peer recommendation usually outranks content; a trusted community thread outranks your homepage. This ordains where Referral and Shareability matter more than advertising.
Map the decision, including its politics. For each buyer persona: what evidence moves them (trial, case study, security documentation, a colleague’s word), what disqualifies instantly, and who else must say yes. Then supply the champion with what they need to sell internally—the buying journey often ends in a meeting you are not in.
De-risk rather than pressure. The honest levers at the decision point are risk reversal and proof: trials (Sandbox Experience), transparent pricing, easy exit (Graceful Exit—leavability is a purchase criterion), and truthful social proof. Manufactured urgency at the moment of peak doubt is the signature dark pattern of this stage (User Agency).

For builders and agents

VoC is infrastructure: pipe support tickets, sales-call transcripts, and review text into one searchable place, tagged by segment and need state. Copy decisions, naming decisions, and search keywords should be queries against this corpus, not brainstorms. Name features, navigation, events, and documentation headings in customer vocabulary—users search your docs and UI with the same words they typed into Google; internal jargon in a menu is a findability bug. (Increasingly, AI assistants answer from your docs: docs written in customer language are how you appear in those answers correctly.)

Preserve arrival context through the funnel: the campaign, query, or referrer that brought a signup should shape first-run and appear in analytics cohorts (Deep-link). Instrument buying surfaces: pricing-page dwell and return visits, checkout abandonment step, trial-to-paid lag. Long lags and repeat pricing visits are doubt made visible—the fix is usually missing reassurance, not a bigger discount. For agents: before rewriting customer-facing copy, ask for (or mine) the VoC corpus; before reviewing a checkout or paywall, ask who the buyer is, what they must verify, and who else has to approve. Persuasion generated without that context defaults to pressure.

When to use it

Before writing homepage, onboarding, or paywall copy
Choosing where Referral and Shareability matter more than ads
Mapping B2B evidence packs (security, case study, champion kit)
Instrumenting pricing and checkout doubt

Do

Mine language before writing; cluster by need state
Reconstruct the last real search in interviews
Map the decision and its politics; equip the champion
De-risk at the decision: trial, transparent pricing, easy exit, truthful proof
Name UI and docs in customer vocabulary; preserve arrival context

Don’t

Brainstorm taglines without a VoC corpus
Manufacture urgency at the moment of peak doubt
Strip campaign/referrer context at signup
Equip a champion with nothing they can use in the meeting you are not in

Founder Tip

If your category jargon and their words diverge, close the gap in their favour—recognition beats education at the first glance.

Make It Yours

Harvest 20 verbatims — cluster by need state; pick phrases for headline tests.
Last search story — for three ICP interviews.
Buyer evidence list — what moves them; what disqualifies instantly.
Doubt instruments — pricing returns, checkout step abandon, trial-to-paid lag.

Insights & Metrics

VoC reuse — Shipped headlines/CTAs that reuse a harvested phrase ÷ New customer-facing strings.
Search–arrival fit — Sessions where landing matches inferred awareness from channel.
Buy doubt signals — Repeat pricing visits and checkout abandon (fix is usually reassurance, not discount).

Behind the Data

Would a customer recognise themselves in your homepage in under five seconds?
Where does peer trust outrank your content?
What burned them last time they bought something like this?

Friction, Calibrated Trust, User Agency, Jobs-to-be-Done
Consumed by: Monetisation, Conversion Optimisation, Growth & Viral, Trust Building, JTBD Copywriting, Product Voice, The Paywall

Agent skill

Primary command: /productfeeling tone — calibrate voice and microcopy to VoC
Related commands: /productfeeling jobs, /productfeeling trust, /productfeeling anti-patterns, /productfeeling sequence (Customer discovery)
When the agent should load this page: “voice of customer”, “copy language”, “how they buy”, “watering hole”, “checkout doubt”
Companion handoff: DocSlime — language and voice in docs/PRODUCT.md / DESIGN.md; buy/competitive notes in docs/strategy/. Impeccable for craft after brief. RedTeam before coercive urgency on paywalls. No external discovery skill.
Feeling north star this practice serves: recognition in their words; trust at the decision
Anti-goals: jargon walls, urgency theatre, stripped arrival context
Reference path: skill/reference/tone.md
Durable DocSlime targets: docs/PRODUCT.md (voice/language); docs/strategy/ (buy journey, competitive)