Product.

The Product Operating Review

5 min read·Apply by lunch

The question

If someone stopped you on the way in and asked what your product learnt last month, what you shipped because of it, what that shipping actually moved, and what the whole thing now costs to keep alive — how many of the four could you answer with a number rather than a story?

The idea

Andrew Grove was Intel's president when High Output Management appeared in 1983, and became its chief executive four years later. Most of the book is about instruments. Grove's picture of production is a black box, and the manager's job is to cut windows into it — because outputs only tell you what has already happened, and by then you can't do much about it. What you want are the few indicators that show you the inputs while there's still time to act. He was just as firm about what happens once people know they're being watched. Any indicator counting quantity will eventually be met by quietly sacrificing quality, so every quantity indicator has to travel with a paired one measuring the quality of that same work, and you read both or neither. And you read them to a schedule rather than when something goes wrong, because a number means nothing except next to the same number from last month. A product review is that discipline pointed at one product: four questions, every count paired, inputs read before outputs. The order isn't housekeeping either. Delivery is the only one of the four columns that keeps its own records, so whichever column you open on decides how the month feels.

The framework

01
Demand first, delivery second — and the order isn't a preference.
Open with what the month taught you about the people you serve: the conversations you ran, the switch stories you heard, the request that keeps arriving and keeps not getting built. Delivery comes second, and the reason isn't tidiness. Delivery is the only one of the four that keeps records by itself — tickets close, releases tag themselves, the board empties on its own. So a month opened on shipping always has something to show, and it reads as a good month right up until the column beneath it comes back flat. Put the column with no automatic trail first, while you've still got the patience to fill it in.
02
The corner that gets cut to make a number look good.
Grove's instruction is to name, for every quantity you intend to review, the quality that could be sacrificed to hit it — and then put both on the page. Features shipped travels with defects reaching customers. Cycle time travels with how much work came back. Count the experiments you ran next to the number you let reach their stop date rather than calling early, and count the customer conversations next to how many were about the customer's life rather than your idea. This isn't bureaucracy. Anything reviewed every month is something people start optimising for by the second month, and a lone number doesn't survive that.
03
Every ship gets one number, agreed before the month starts.
This is the column that keeps the review honest, and its rules have to be set in advance or they'll be written to fit the result. For each thing you intend to ship, write down the one measure it's meant to move and roughly how far. Then report what happened, including the ones that moved nothing, which will be most of them. Read retention by cohort rather than in total, because a rising total will happily hide a curve that never flattens. Split the month itself four ways underneath — stayed, new, resurrected, churned — and say whether arrivals still outrun departures; the fit score sits here too, quarterly rather than monthly, since that one only reads against a fixed sampling rule. And leave the failures on the page: a review that only ever carries wins can't be used to decide anything.
04
What is it costing you to keep last year's product alive?
Nothing on a roadmap is ever scheduled for removal, so the liability only accumulates — the features nobody opened this month that everybody's code still has to support, the support load each one drags behind it, the shortcut from three releases back that's now taxing every change near it. None of that reports itself either, which is exactly why it belongs on the same page as the wins. Give it four lines, and read them before you agree to build anything large. A product that costs more to keep than to extend has already made your roadmap decision for you.

Members only · 4 principles + template + AI mentor

364 more frameworks are waiting.

The Product Operating Review is Day 1 of 365. One framework every morning for a year — across strategy, sales, negotiation, leadership, and more.

$1 /day

Billed $365/yr · cancel any time

365 frameworks

12 topics · templates · AI mentor

Reserve your place — $365/yr →

One payment a year · the whole library · no auto-renewal.

More Product frameworks

All 22 product frameworks →