Skip to content
demandboxskills
All skills
Research

Voice of Customer

Mine transcripts, reviews and threads into the language dictionary that feeds every asset.

Version
v2026.10.1
Updated
Oct 24, 2026
Size
7 KB

What you get

Pull the exact words buyers use from Gong and Fireflies transcripts, discovery calls, G2 and Capterra reviews, Reddit and Quora threads, support tickets and win/loss notes, and turn them into a customer-language dictionary for ads, pages, emails and positioning. The upstream skill that makes every downstream skill sharper.

Skill content

DemandBox — Voice of Customer

Most B2B copy fails because it sounds like the marketing team wrote it. Winning copy sounds like the buyer wrote it. This skill builds the raw-material pipeline that makes that possible — and then packages it into a form the copy-writing, ad, and LP skills can pull from.

The single rule

Do not invent customer language. Lift it, verbatim, from the source. Every headline, subject line, hero claim, ad hook, and objection handler on a page should be traceable back to a real customer utterance.

Sources, ranked by signal quality

Tier 1 — direct, unfiltered

  1. Sales discovery call transcripts (Gong, Fireflies, Granola, Chorus). The prospect describes their pain in their own words before they know what they're being sold. Gold.
  2. Win/loss interviews — what made them choose us / the competitor / to stay with status quo.
  3. Churn interviews — the clearest signal of the promise we didn't keep.
  4. Support tickets and customer complaints — active friction in their own words.
  5. Direct outbound replies (positive and negative) — compressed, emotional, real.

Tier 2 — public but still direct

  1. G2 / Capterra / TrustRadius reviews — pros, cons, "what problems are you solving," "what do you dislike." Filter by ICP-fit firmographics.
  2. Reddit threads in target subs — high-trust, peer-to-peer, unfiltered.
  3. Quora and LinkedIn comments on relevant topics.
  4. Indie Hackers / HackerNews discussions for technical audiences.
  5. Product Hunt comments on competitor launches.

Tier 3 — adjacent, lower signal but high volume

  1. Newsletter replies (if the client has a newsletter).
  2. Community posts (Slack / Discord / Circle groups in the category).
  3. Podcast interview transcripts with buyers in the ICP.
  4. Sales call recordings from competitors if obtainable legally (public webinars, analyst calls, earnings transcripts for public companies).

Extraction method

For each source, extract five things:

  1. Pain statements — "I can't get my sales team to respond to MQLs within 15 minutes."
  2. Outcome statements — "I want a single dashboard where finance sees real CAC."

Sign up free to read the whole skill and download it

We'll email you a sign-in link and bring you right back here.

No password. We email you a single-use link.

Version history

  1. v2026.10.1 · Oct 24, 2026

    First public release.