The Minimal Marketable Product¶
Definition¶
Because no market-research technique forecasts customer response with certainty, the best available strategy is to envision the minimal marketable product (MMP) — "a product with minimum functionality that meets the selected customer needs" — rather than trying to specify a complete, correct product upfront. The term, which Pichler credits to Denne and Cleland-Huang's "minimal marketable feature set," is deliberately narrower than a minimum viable product built purely to learn: an MMP is meant to actually sell, not just to test a hypothesis.
In the Book¶
The chapter pairs two contrasting Apple stories. The original iPhone (2007) shipped without copy-and-paste, group text messaging, or a software development kit — features standard on competing phones — because Apple selected a narrow set of customer needs rather than matching every competitor feature, and the limitations didn't stop its success. The Apple Newton (1993), by contrast, launched after five years of development trying to do everything, including handwriting recognition that didn't work; it was withdrawn from the market in 1998. A third case, Expertcity's 1999 interactive technical-support product, shows the adaptive payoff of shipping small: when the product underperformed, the company noticed users repurposing one feature — a desktop-sharing utility — for remote computer administration, pivoted the product into GoToMyPC, and was acquired by Citrix for $225 million in 2003. The book credits the minimal-product strategy with faster time to market, lower cost, earlier cash flow, and faster learning — and quotes Google's Marissa Mayer on deliberately expecting to throw out many products in the process.
Why It Matters¶
When prediction is unreliable — which is most of the time in any novel or competitive undertaking — betting big on a fully-specified plan multiplies the cost of being wrong. Committing to the smallest version that could still succeed converts an expensive, one-shot prediction problem into a cheap, repeatable learning loop, with the option to adapt (as Expertcity did) once real feedback arrives.