06-reference/research

us household printer access printable subscription tam

2026-09-16·research-brief·source: deep-research·by Ray Data Co (deep-research synthesis)
scribble-workstammarket-sizingprinter-accesswillingness-to-pay

Printer access can be narrowed and it moves up; willingness-to-subscribe cannot be measured but every benchmark says 15–40% is 3–5x too high — and the two corrections move the TAM band down one full notch

The question

"What share of US households with kids 3–10 have convenient home-printer access and would actually subscribe to a printable-games product, replacing the unmeasured 40–60%/15–40% assumption ranges driving Scribble Works' $14M–$110M TAM estimate?"

Context: [[2026-09-11-business-opportunity-pricing-and-acquisition]] builds a three-factor funnel off a 21.3m-household demographic base and labels all three narrowing factors "unmeasured scenarios." This brief attacks two of the three. It does not attack the demographic base or the $9/month price point.

What we already know (from the vault)

What the web says

Convergences and contradictions

Synthesis for RDCO

One of the two factors can be narrowed with evidence; the other cannot, and pretending otherwise would be the failure mode. Printer access moves, and it moves up. The model guessed 40–60% for a segment that two independent-incentive sources bracket at 62–75%+ ownership. But the model's variable is not ownership, it is convenient access, and no source measures that. So the correction is a decomposition rather than a replacement: P(convenient access) = P(owns a printer) × P(that printer is working, inked and low-friction enough to use monthly). The first term is now evidence-informed at 60–80%. The second term is pure inference — I am assigning it 0.55–0.95 and that is my judgment, not a measurement, anchored only on the direction of the WithMe apartment data and the founder's own 1-in-4 cohort design. Multiplied out: 35–75% convenient access, midpoint 55%. Midpoint barely moves. The band gets wider. The 40–60% range in the vault was falsely precise, and it was falsely precise because it silently collapsed two factors into one.

Willingness-to-subscribe cannot be narrowed with evidence, and the evidence that exists says the current range is 3–5x too high. No operator publishes a printables-membership conversion rate. What is available is a dense set of adjacent freemium benchmarks that all land between 1% and 15%, clustering 2–8%, against a model range of 15–40%. The obvious defense is that the model applies its 15–40% to an already-qualified denominator (households that both print conveniently and have recurring need), not to raw free-signup traffic — that defense is legitimate and it is why I do not cut all the way to 2%. But "recurring need" is itself a 10–20% guess, so the qualification is asserted rather than demonstrated, and stacking an unmeasured qualifier to justify an above-benchmark conversion rate is circular. Recalibrated: 3–15%, midpoint 7%, sitting on daydream's high-intent band with room below it for the possibility that the need filter is softer than assumed. I left the middle factor, recurring need 10–20%, completely untouched — I found no evidence bearing on it in either direction, and moving it would be fabrication.

Run through the model's own arithmetic, the whole distribution slides down almost exactly one notch, and the coincidence is worth stating precisely. 21.3m households × the revised factors × $108/year gives a floor of 22,365 households / $2.4m, a midpoint of 123,008 / $13.3m, and a ceiling of 479,250 / $51.8m. Set that against the current $13.8m / $51.8m / $110.4m: the revised ceiling is the current midpoint to the dollar, and the revised midpoint is within 4% of the current floor. The band also widens from a 7.9x multiplicative spread to 22x, which is the honest cost of un-collapsing the printer factor. Anyone who reads "$14M–$110M" and hears "call it fifty" should now read "$2M–$52M" and hear "call it thirteen, and the error bars still cover an order of magnitude."

The decision this actually forces is to stop refining the TAM, because at every point in the revised band the TAM is not the binding constraint. The studio charter's break-even is 240–970 subscribers ([[2026-08-31-studio-charter]]). Against the revised floor of 22,365 households, 971 subscribers is 4.3% of the entire national market — genuinely hard for a solo operator, and the one scenario where market size bites. Against the revised midpoint it is 0.79%, and against the ceiling 0.20%. Three of those four numbers are rounding error. That asymmetry means further TAM work has almost no decision value: the band cannot produce a kill, and it can only produce a go that the founder already has. What would move the estimate is the instrument that is built, staffed, and sitting unrun — the cohort week-1 kit, 0 of 4 households, already designed to capture printer access at hand-off. Four households will not give a confidence interval. But four households produce the first real observation on both factors at once, and the current state of the evidence is that a fifth hour of desk research on printer penetration produces less than one household producing one reprint.

Why this is in the vault

It replaces two named "unmeasured scenarios" in the sizing table of [[2026-09-11-business-opportunity-pricing-and-acquisition]] with an evidence-informed band for one and a benchmark-anchored haircut for the other, and it revises the headline $14M–$110M figure to $2.4M–$51.8M. Concretely: it stops the $51.7m middle case from being quoted as the expected outcome when it is now the optimistic ceiling, and it supplies the argument for un-parking the cohort week-1 test as the cheapest available instrument on the same two variables.

Open follow-ups

Related

Sources

Vault

Web (all fetched or searched 2026-09-16)

Not carried forward: a "92 million active home printer installations in the US" figure appeared in search-result summary and could not be confirmed in any fetched page. Recorded as a non-fact so it is not rediscovered as evidence.