For a two-person startup, the useful measure of an AI marketing function is the time between supplying accurate product context and receiving a draft a founder can review. Content volume can rise while the real bottleneck, founder attention, stays exactly where it was.
In April 1970, the Apollo 13 crew faced rising carbon dioxide inside the lunar module Aquarius. The command module had square lithium hydroxide canisters, while the lunar module used round openings. The parts available in space did not fit the system keeping Jim Lovell, Jack Swigert, and Fred Haise alive.
At NASA’s Manned Spacecraft Center in Houston, engineer Ed Smylie and his colleagues had to produce instructions the astronauts could follow using materials already aboard the spacecraft. The standard that mattered was brutally practical: how quickly could the team turn a precise problem and a fixed inventory into a workable procedure that Mission Control could check, transmit, and use?
NASA’s historical accounts document the improvised adapter, commonly called the “mailbox,” built from items including plastic bags, cardboard, a hose, and duct tape. It worked. The crew returned safely.
A startup’s marketing stakes are smaller, but the operating lesson holds. Count the time required to turn known product facts into something a responsible person can inspect. That interval reveals whether the system removes work or merely produces more of it.
Why content volume hides the real constraint
A dashboard can report 30 posts, eight articles, and five videos. None of those numbers tells you how long the founders spent correcting the target customer, removing invented claims, reconciling pricing, or translating one generic idea across channels.
For a two-person company, review capacity is scarce. One founder may be shipping a release while the other handles support, sales calls, and billing. A queue full of drafts creates another inbox. If each item needs fifteen minutes of reconstruction before anyone can judge it, higher output increases the burden.
Time to first reviewable draft measures a more useful chain:
Product context enters the system. The system applies the defined audience, positioning, offer, evidence, voice, and channel requirements. A draft arrives with enough accuracy and structure for a founder to approve, edit, or reject it.
“Reviewable” carries most of the weight. A reviewable draft has a clear argument, reflects the current product, avoids unsupported proof, and fits its destination. It does not need to be perfect. It needs to be close enough that the founder is making a marketing judgment instead of repairing missing context.
That distinction also matters when three conflicting notes must become one defensible draft. Fast generation means little when the underlying inputs disagree.
Establish the clock before adding automation
Start the measurement when the required context is available, not when someone clicks “generate.” Otherwise, the metric rewards systems that hide preparation work in another tab, document, or person.
Stop the clock when the first complete draft is ready for human review. Record three times separately:
- Context preparation time covers updating product facts, audience, pricing, proof, and constraints.
- Generation time covers research, angle selection, drafting, and channel adaptation.
- Review time covers checking claims, voice, links, screenshots, platform fit, and approval boundaries.
This breakdown shows where the function is actually slow. If context preparation takes two hours, the product strategy may be scattered. If generation is quick but review takes an afternoon, the drafts may lack grounding. If both are short but nothing gets published, approval or provider access is the constraint.
Marketing Agent gives each product a defined target market, ideal customer profile, positioning, brand voice, competitors, keywords, pricing, offers, and channel strategy. It can use that shared context to generate blogs, social posts, newsletters, technical articles, and videos, then publish approved or unattended work where connected providers permit. The relevant promise is continuity from strategy to review, with product-scoped permissions and approvals kept visible.
Run a one-week measurement change
For one week, stop celebrating raw draft counts. Choose one recurring job, such as a weekly technical article with supporting LinkedIn and X posts.
On Monday, record the time spent bringing the product context current. Include changed pricing, newly shipped capabilities, available customer evidence, and any claims that remain unknown.
For every content cycle, capture the start time and the moment the first complete package becomes reviewable. Then record founder review minutes and the reason for every substantial correction. Use plain categories: wrong audience, stale feature, unsupported claim, weak angle, voice mismatch, or channel mismatch.
At the end of the week, inspect the correction log. Repeated errors should become changes to the shared product strategy, source material, or approval rules. This turns review feedback into an input for the next cycle instead of a recurring editing tax.
Do not lower the bar by calling rough output reviewable. A pile of plausible paragraphs can arrive in seconds and still take an hour to make trustworthy. The goal is a shorter path to an informed decision.
Make the metric improve the whole function
Once the first reviewable draft arrives faster, measure what happens next: approval time, publication rate, corrections after publication, engagement supported by available analytics, and whether performance informs future angles. Speed without learning can produce the same weak idea more often.
Keep volume as a capacity measure. Pair it with the time metric and a quality check. A practical weekly view might show median time to first reviewable draft, founder review minutes per asset, percentage approved without structural revision, and the top correction reason.
This also guards against uncontrolled automation. Unattended scheduling can run multiple times per day, but only after the product context, permissions, schedules, safety caps, and capability limits deserve that trust. Faster review should help the founders decide what can run automatically and what still needs approval.
Apollo 13’s “mailbox” mattered because Mission Control converted exact constraints into a procedure the crew could evaluate and execute. Your marketing function should do the same on ordinary weeks: take the product you actually built, the evidence you actually have, and the channels you actually control, then produce a draft worth reviewing before the next founder task takes over.
Comments
No comments yet.