← Back to cases
CASE A03  /  CONTENT & SEO · AGENCY

Brief-to-first-draft in 24 hours, with brand voice QA built in.

A 28-person content & SEO agency in Amsterdam was burning 60–90 hours per month on the same brief-intake cycle: incomplete client briefs → 3 days of clarifying questions → late drafts → revisions because tone-of-voice missed. We mapped one workflow end-to-end and built an intake form that nudges clients toward completeness, an AI assistant that drafts the first version against the client brand voice, and a QA pass that flags tone, claim and disclosure issues before review.

−65%REVISION CYCLES
4 → 1.5ROUNDS / PIECE
+40%ON-TIME RATE
5 wkTO ROI
Writer working on a laptop with notebook beside
CONTENT AGENCY · AMSTERDAM · MAR 2026 · CONFIDENTIAL DETAILS REDACTED
THE PROBLEM

Brief comes in. Three days of clarifications. Draft is late. Tone misses. Repeat.

The agency was running ~80 content pieces per month across 14 clients. Average revision cycles per piece: 4. The pattern was always the same: incomplete brief → email back-and-forth → late kickoff → draft written against an outdated tone-of-voice doc → client comes back asking 'why does this sound like a different agency wrote it?'

Senior editors were spending 50% of their time on tone fixes that should have been caught at draft stage. The agency hit a ceiling on capacity — they couldn't onboard a 15th client without hiring two more editors.

Every draft was a guess at what the client really meant. Then the client told us. Three drafts later.
THE PROTOTYPE

Smart intake. Per-client brand voice. Pre-review QA on every draft.

Three components, one workflow. (1) An intake form that asks 'what would a great brief have' for each client type — and nudges the client toward filling the gaps before submission. (2) A per-client brand voice model: distilled tone, banned phrases, must-use terms, claim/disclosure rules, all sourced from their brand book and 20+ approved past pieces. (3) An automated QA pass that runs on every draft before editor review, flagging tone drift, missing disclosures, banned claims.

The draft itself is still written by the writer, often AI-assisted. But the writer is no longer guessing at brand voice — the rules are explicit and the QA pass is non-negotiable before submission.

THE PILOT

Six weeks across 3 clients with the worst revision pattern.

We piloted with the 3 clients who had the highest revision cycles. By week 4, average rounds per piece dropped from 4.1 to 1.8. By week 6, on-time delivery rate moved from 60% to 86%. Editors reported they were doing real editing again — substance, structure, argument — not tone policing.

We did more billable strategy work in 6 weeks than in the previous quarter. Same headcount.
THE OUTCOME

Eight weeks after full rollout across 14 clients.

  • Average revision cycles per piece−63%4.1 → 1.5
  • On-time delivery rate+26 pp60% → 86%
  • Editor time on tone fixes−72%Freed for substantive editing
  • Client capacity without hiring+25%14 → 17 retainers
  • Time to first measurable impact5 weeks
WHAT'S NEXT

Extending the QA pass into client-side claim approval and compliance routing.

Phase 2 routed flagged claims (regulated industries — finance, health) into a compliance approval queue with audit trail. Phase 3 will tackle the same pattern for video scripts and short-form social.

YOUR CONTENT WORKFLOW

Revision cycles climbing and editors editing tone instead of substance?

20-min audit. Bring one client and one recent piece that took too many rounds. We will reverse-engineer where the cycles got eaten and where AI-assisted QA actually helps.

Take 2-min assessment