Leadership1 publisher3 min readPublished
Itamar Gilad gives a good product idea at best a 20% chance of surviving prioritization
Itamar Gilad estimates that at best 20% of good product ideas survive prioritization in a launch-and-iterate process, and he calls the figure very speculative. His case puts most of the loss in the planning meeting, where consensus and rank decide what gets built.
The Board Room · Leadership desk
What happened
- In launch-and-iterate, prioritization is a set of binary calls: a small minority of ideas are approved for full builds and most are parked or killed.
- Gilad estimates, calling it very speculative, that a good idea has at best a 20% chance of surviving the prioritization round.
- He argues that committee review at the specification stage can catch design flaws but can also produce a mediocre compromise.
Compiled by The Board RoomSomething wrong?How this is made
Why it matters
- decision Who sits in the planning forum becomes a product decision in its own right, since in Gilad's account an unbacked idea loses to executives' pet projects and Sales demands.
- exposure Novel, counterintuitive or unimpressive ideas are the ones most likely to be rejected, so a roadmap chosen by consensus leans toward familiar work.
- constraint The essay gives a figure for only the first of three phases, so a leader cannot use it to size the total loss through specification and implementation.
- cost Testing first moves team effort onto ideas that will mostly fail, and a leader pays that cost to avoid losing the rare good idea at the roadmap meeting.
Gilad's argument starts from a base rate. He cites A/B-test research showing that the vast majority of ideas deliver little or no value, and that some detract value [1]. He says a product's feature usage and past roadmaps show a power law, with a small minority of ideas creating significant impact, because most of the low-hanging fruit has already been picked [2]. His question to leaders follows from that scarcity: "is your product process designed to find and cultivate these rare good ideas?" [11]
In a launch-and-iterate process the first sorting step is prioritization. Gilad describes it as a set of binary decisions in which a small minority of ideas are approved to be built in full and most are parked or killed, usually on the basis of consensus, opinions and rank [3]. He argues that the uncertainty pushes people toward shortcuts such as risk aversion, halo effects and groupthink. Some of the best ideas are novel, counterintuitive or unimpressive, he writes, so they may not do well in that room [4]. Politics adds a second filter. A good idea competes with executives' pet projects, customer requests, Sales' "deal-closing" demands and team favourites [5]. "Many good ideas die just because they don't have the right backer," he wrote [6].
The board-deck version of the essay is one number. "This is very speculative, but I'd estimate that the odds of survival of a good idea during prioritization are at best 20%," Gilad wrote [7]. Taken as a ceiling, that means at least four of every five good ideas are gone before anyone writes a spec [1]. That version is incomplete in two ways. Gilad labels the figure a guess, and it covers only the first of three phases: prioritization, specification and implementation [3].
The losses continue after approval. Gilad argues that only a small subset of the ways to build an idea will be valuable, usable, feasible and business-viable, the four risks he credits to Cagan. One wrong turn in the interface, or one removed capability, can cost an idea much of its impact [8]. Stakeholder review can catch bad assumptions and design flaws, he writes, but it can also produce bad product decisions or a mediocre compromise [9]. "Does the process of decision-by-committee help surface the best implementations? Sadly no," he wrote [10].
The central figure is Gilad's own estimate, and it has no data under it. The base-rate claim rests on A/B-test research [1]; the 20% rests on the author's judgement [7]. In my view the direction of the argument survives a wrong number. A process that approves ideas in full on consensus and rank gives an untested idea one hearing, in a meeting, against sponsored projects [3][5]. The essay sets out to compare that model with evidence-guided development and to calculate a good idea's odds under both [12]. The text available ends in the specification section, before those figures appear [13].
Testing ideas before committing to them spends team time on candidates that will mostly fail, because most ideas deliver little value [1]. Launch-and-iterate saves that time and spends it on full builds chosen by consensus and rank [3]. The decision falls in the next planning cycle. Whoever sits in that forum decides which ideas get tested and which get built, and under Gilad's estimate the forum as usually run keeps at most one good idea in five [7][1].
What to watch
- Gilad's own survival estimates for the specification and implementation phases, which would show whether prioritization is the largest loss or only the first.
- His survival odds for the evidence-guided approach; if they are not much better than 20%, the case for testing first weakens.
- A citation for the A/B-test research behind the base rate, which would let leaders check the premise the whole argument rests on.