A real week in the life. The work that actually pays back the $20/seat, and the work that doesn't.
Most teams use Claude for the wrong things. They write tweets in it. They summarise meetings in it. They paraphrase blog posts in it. None of that pays back the $20/seat. The work that does (the reporting, the briefing, the strategic synthesis) gets done in spreadsheets and Google Docs like it has been for a decade.
This is what a six-person marketing team I worked with actually does, week to week, after we re-routed their Claude usage. The time recovered came back at about 4-5 hours per marketer per week. The work that moved the needle didn't change. The work that quietly drained the calendar did.
If you're using Claude primarily for short-form content (tweets, social posts, quick paraphrasing) you're using it for the work where the output quality matters least. The marginal value of a slightly better tweet is small.
The work where Claude's value compounds is the work that takes hours and is procrastination-shaped: the weekly readout, the campaign brief, the competitive synthesis, the content brief. Move Claude there and the $20 pays back ten times.
Before the week-in-the-life, three things have to be in place:
One hour of setup. Compounds for the next 90 days. We're now in week 6.
9:00. Data export. Whoever owns analytics exports the week's GA4 + ad platform + email numbers into the shared sheet. 30 minutes.
9:30. Run the Performance Readout Narrator. Paste the data into the Project, add this week's goals and the previous week's readout. Claude returns: headline, three findings ranked by decision-relevance, two follow-up questions, one recommendation per finding. 5 minutes.
9:35. Human review. Read the readout. Check the data for the bits Claude flagged as needing verification. Edit for nuance. The "refuse to declare a trend from fewer than two data points" instinct is baked into the prompt, so the readout doesn't manufacture a story, but you still need to make sure the framing matches the team's reality. 30 minutes.
10:05. Distribute. Paste the readout into the team standup doc + a 3-bullet summary to the leadership Slack. 5 minutes.
What changed: The readout used to take 4 hours of an analyst's morning. The first draft now takes 5 minutes. The 30 minutes of human review is genuine analyst work: the rest is mechanical synthesis Claude does well.
For each piece of content the team is shipping that week:
Gather the inputs (target query, top 10 SERP excerpts, site context). 15 minutes per brief.
Run the SERP-Informed Brief Writer. 5 minutes per brief.
Spot-check the consensus structure claims, pressure-test the content gaps. 15 minutes per brief.
What changed: Briefs used to take 90 minutes each, mostly because the writer had to do their own SERP research. Now the SERP research is structured by Claude and the writer can spend their time on the actual writing. We do 2-3 briefs on Tuesdays instead of 1.
Not every Wednesday. About every third week, when a campaign is in flight.
The team lead opens a Claude Project conversation. The Campaign Brief Generator prompt is pinned in the Project Instructions. The lead chats through the campaign. Claude asks the questions, refuses to output the brief until every required section has real content.
Output: a campaign brief with all the required sections, objective, audience, KPIs, channels, creative direction, measurement plan, non-goals. The lead reviews, edits, ships to the writer/designer.
What changed: Briefs that used to take a half-day to write now take 45 minutes. More importantly, briefs now always have a measurement plan and a non-goals section, because Claude refused to output without them.
Thursday is no-meetings day for this team. Deep work, no synchronous interruption.
Claude usage on Thursdays is whatever's needed for the deep work in progress. Writing a pillar article? Voice guide is pinned, draft happens in Claude Artifacts. Doing competitive synthesis? Run the Teardown prompt and spend the rest of the day on the positioning conversation. Building a new email sequence? Use the lifecycle prompts (paid pack) to map the flow.
This is the day Claude's value is hardest to measure, but it's also the day the team ships most of what compounds.
Competitive monitor: Whoever owns competitive intel grabs the homepage and pricing page text from each tracked competitor (5-7 of them). Pastes into the Project. Asks "what changed vs. last week?" Outputs a 3-bullet Slack message. 30 minutes.
Agent triage: Quick review of how the week's agents performed. Are the auto-drafts in Gmail useful? Did the content repurposer produce on-voice output? Anything that needs the system prompt iterating? 30 minutes.
End of week. Standup-doc gets one final read. Slack threads close out.
Time recovered per marketer per week: about 4-5 hours. That's 200+ hours per marketer per year. For a six-person team, ~1,200 hours per year: the equivalent of hiring half a marketer.
Cost of Claude across the team: $120/month ($20 × 6 seats). Annualised: $1,440. Versus the value of 1,200 hours of senior marketing time: roughly $90,000 at fully-loaded cost. ROI is not the right framing here, it's "this is too low-cost to skip."
The team doesn't use Claude for tweets. They write those by hand, because the marginal value of a slightly better tweet is small and the voice quality matters more than the speed. They use Claude for the readouts and the briefs and the synthesis (the work that takes hours and is procrastination-shaped) and that's where the calendar comes back.
Four moves. About six hours of setup. Pays back four to five hours every week, every week. It's not magic, it's just routing the right work to the right tool.
All 30 prompts, 6 playbooks, 5 agent blueprints, the Reporting Kit, quarterly updates. $549/year flat for the whole team, no per-seat fees. 30-day no-questions refund.
See the Suite →