# Playbook: VoC Mining — Voice-of-Customer Assembly

> Mine reviews, support tickets, and interviews for the customer's own words; the copy is assembled, not invented.

## Origin

**Joanna Wiebe**, founder of **Copyhackers**. Wiebe's contrarian claim against the entire "clever headline" tradition: the best line on the page was very often already said, verbatim, by an actual customer in a review, a support ticket, or an interview — the copywriter's job is to find it, not out-write it. This is a direct rebuttal to the idea that persuasion is manufactured; it's the practical extension of Schwartz's keystone belief (§2.1) that desire is pre-existing — Wiebe extends it one step further: not just the *desire* but the exact *phrasing* of that desire already exists, out in the world, waiting to be collected.

## When to use it

- **Whenever real customer language exists** — per §5.2's decision table, this applies "regardless of other choices": VoC mining runs *before* drafting even when another lead type or school (RMBC, PAS, PASTOR) governs the final shape.
- For 19Keys specifically: Sovereign CRM notes, IG/ManyChat DM threads, testimonials, comment sections — all named explicitly in §6.6 as available source material.
- SaaS, considered B2C/B2B, and any market with an existing installed base of users who've already articulated the problem or the win in their own words.

## When NOT to use it

- **Brand-new categories with no existing customers to mine** (§4's explicit failure case) — an Unaware-market, first-to-market launch has no verbatim corpus yet; fall back to Schwartz-grid reasoning and the founder's own articulation of the problem until real usage data accumulates.
- As a substitute for judgment — VoC supplies raw material and validation, not structure. It still needs to be poured into a lead type (Great Leads) and a spine (§3.2); a page that's just a pile of quotes with no architecture fails the same way any unstructured page does.

## Step-by-step build

1. **Collect a swipe file of verbatim phrases** — export or transcribe directly from source; no paraphrasing at this stage, even if the customer's phrasing is grammatically rough. The roughness is often the credibility signal (Ogilvy: the reader is not a moron, and neither is the reviewer).
2. **Cluster by desire and by objection** — sort phrases into "what they wanted," "what almost stopped them," "what convinced them," "how they describe the result now."
3. **Rank by frequency and specificity** — a phrase five different customers land on independently is a stronger headline candidate than a single eloquent outlier; a phrase with a concrete number or scene beats a generic compliment.
4. **Build headlines and bullets from the strongest verbatims** — light editing for length only; preserve the customer's actual words and word order where possible, since that's the fluency-as-truth mechanism (§2.6) working in your favor — a stranger's cadence, not a copywriter's, reads as unrehearsed and therefore true.
5. **Feed the objection cluster directly into the objection-handling block** (§3.2.6, minimum three) — VoC-mined objections are stronger than invented ones because they're verified live doubts, not hypothetical ones.

## Worked mini-example

**Raw DM excerpt (anonymized, paraphrased for this example):** "I almost didn't book because I've paid for 3 different 'proof of concept' things before and none of them meant anything to an actual buyer."

**Mined objection:** "This is just another portfolio site that means nothing to a real buyer."

**Headline built from the cluster (not this single quote, but the pattern across several):** "The readiness signal buyers actually check — not another portfolio nobody underwrites."

**Objection-handling copy drawn straight from the cluster:** "If you've built a 'proof of concept' before that a buyer looked at and shrugged — this isn't that. The Ladder score is the thing a buyer's team runs before a call, not after."

## How the verify gate applies

VoC-mined copy still has to clear the standard §8.1/§8.5 rubrics — a verbatim-derived headline still counts against the ≤14-word cap on hero headlines, and a mined objection still counts toward (and helps satisfy) the ≥3 objection-handling minimum. Where VoC mining most directly helps the gate: the **banned-vocabulary/hype scan** (§8.6) is far easier to pass when the language is a customer's own plain words rather than a copywriter's superlatives — hype is rare in real complaints and reviews. Caution on the **proof-coverage gate**: a customer's verbatim number ("saved me 10 hours a week") is still a numeric claim and still needs a tagged source (`[testimonial]` at minimum, ideally corroborated by `[data]`) — VoC origin does not exempt a claim from being tagged.