The team is shipping more. Your CMO can see the cadence shift; your finance team can see the line item for AI tooling. The bridge between the two is a number, and “we feel faster” is not the number. Your finance team wants drafts per writer per week, time-to-publish, and cost per qualified piece — three metrics with named calculations and disclosed assumptions.
Drafts per writer per week
The most useful productivity number, because it is the easiest to defend.
- Numerator: drafts that cleared the editorial gate — not drafts that left the model, drafts that survived self-review and editor review.
- Denominator: full-time-equivalent writer-weeks on the team in the same window.
The trick is the editorial-gate qualifier. A team running raw model output ships fifty drafts a week and lands three; the denominator-corrected figure is the only one a finance team will accept.
Time-to-publish
The cycle-time number. Median, not mean — one rogue six-week piece does not get to skew the figure.
- Window: brief approval to publish.
- Aggregation: rolling four-week median.
- Cut by: content type. A long-form pillar piece and a thread are not on the same clock; one number for both is folklore.
Cost per qualified piece
The number your finance team actually wants. The number that turns “AI tooling helps” into “AI tooling at $X per month produces Y qualified pieces, against the prior baseline of Z.”
- Numerator: writer time × fully-loaded rate, plus editor time × rate, plus monthly tooling cost amortised across the period.
- Denominator: pieces that cleared the editorial gate.
- Comparator: the same calculation against the prior period’s workflow.
The rate trick
Cost figures live or die on the hourly rate. A “97% cost reduction” at one rate is a “70% cost reduction” at another, and the percentage that lands in your CMO’s slide is the one with the rate disclosed.
The clearest worked example is The Content Wrangler’s webinar-description workflow, which moved 100+ webinar descriptions per year from manual writing to a modular prompt library on the Promptitude platform. Annual cost dropped from $25,000 to under $800 — approximately a 97% reduction at a $250-per-hour fully-loaded rate (Promptitude case study, 2024 — vendor-published). About a hundred hours per year reclaimed. At a $50-per-hour rate the same hundred hours cost $5,000, and the headline percentage looks different — the rate has to be in the slide, not in a footnote.
The case clears the case study standards and is logged in the subscriber-library register with the source and the rate caveat carried through, exactly because the rate is the load-bearing assumption. Your CMO’s slide carries the rate the same way.
What this earns you
A productivity story that survives the vendor-procurement conversation and the half-yearly review. The CMO who asked “is the AI tooling actually paying back” closes the question with a number, the unit, the assumption, and a comparator.
The full measurement framework — segment-level cuts, the qualified-piece definition for ten content types, and the dashboard schema — lives in the Premium bundle. The free sample chapter sets up the four jobs the framework measures.