Working Backwards (Problem-First Product Development)
Start from the customer problem, not the ingredients you happen to have in the pantry
- Difficulty
- Moderate
- Time to result
- ~weeks to results
- Steps
- 5
- Confidence
- 95%
Working backwards is not the press release — the press release is a mechanism that forces working backwards. The substance is refusing to process any solution until the customer problem is named and felt. McAllister argues most teams that claim to work backwards are retrofitting: they picked the solution first, then reverse-engineered a problem to justify it.
Origin
The working backwards process is Amazon's, not McAllister's — he is explicit that he 'just used it and just described it' in an early Quora post about Amazon's product development process. He learned it in his first years at Amazon (2006 onward), partly through direct coaching from Jeff Bezos in doc reviews.
Core principles
- 01The problem is the anchor. Everything downstream is negotiable; the problem is not.
- 02The internal press release is a forcing mechanism, not a launch artifact — it is never actually published.
- 03If you can't write the problem paragraph, you don't have a problem.
- 04A compelling problem feels compelling BEFORE you know the solution. That's the test.
- 05Once the muscle is built, the format stops mattering — you can work backwards inside a PRD, a deck, or a conversation.
How to run it
- 1
Write the problem paragraph first — and only the problem
Open the internal press release with a paragraph describing the customer and the problem they have. Do not let a solution enter the document yet.
Pro tip Read the paragraph to someone who knows nothing about your proposed solution. If they don't nod at the problem, stop.
Watch out If the problem paragraph is hard to write or keeps getting skipped, that is diagnostic — there probably isn't a real problem.
- 2
Write the solution paragraph
Now describe the solution, framed strictly as the answer to the problem you just stated. If there are three problems, rank them one, two, three and solve in order.
Pro tip Ask 'what's the most elegant way to solve it?' rather than 'what can we build with what we have?'
- 3
Write the customer quote
Draft the quote a real customer would give about the launched product. It forces you to state the benefit in the customer's language, not yours.
- 4
Write the FAQ — the legitimate plan to succeed
The FAQ is where you prove there is a legitimate plan: the internal components, the finances, the key technical hurdles. It is the least-written-about and most load-bearing part of the process.
Pro tip The FAQ is what earns you the resources — it demonstrates you have been thoughtful enough, given the product's stage, to deserve them.
Watch out A big idea with no FAQ is a wish. Bezos's third test kills it.
- 5
Hold a working backwards review as a gate
Convene leadership to review the concept, the press release, and the FAQ, and to ask questions. Ideally this happens at the outset and functions as the approve/greenlight gate before any team is staffed.
Pro tip Even when it isn't a formal gate, the review is a superb way to ground a wider group in the customer problem before they see the solution.
Watch out Below a certain scale the process has real overhead and isn't worth it. Reserve it for significant or net-new products.
In the wild
McAllister brought a press release for a concept to a review with Bezos. It had no problem paragraph — he'd skipped it. Bezos's response was not to demand a rewrite but to reframe: maybe if you don't have a problem paragraph, there isn't really a problem.
→ McAllister recognised he had had the solution in mind all along and had skipped the working-backwards step entirely. He now says his brain is wired such that he cannot process a new initiative until the problem and customer are on the table.
In McAllister's first years at Amazon, the team was excited about a project Bezos had sketched in a meeting: linking products by subjective attributes and letting customers vote on them. The tell was in the name — 'ASIN to ASIN linking' — an internal identifier, not a customer problem. They wrote the press release anyway and shipped it.
→ It wasn't successful, and McAllister later shut it down. It became his canonical example of using the working backwards process without actually working backwards.
Tasked with increasing direct-traffic loyalty without a paid program like Prime, McAllister worked with no team, profiled several concepts through the working backwards process, and reviewed them with Bezos and Wilke every couple of months.
→ The working backwards review was the greenlight gate — Amazon Smile was the concept that passed it, and only then was a team built and the product launched.
Common mistakes
Retrofitting the problem onto a chosen solution
Teams start with 'these two technologies look similar, we could combine them', then add a customer and a problem afterwards to make the doc look right. Using the template does not make it working backwards.
Treating the press release as the point
The PR/FAQ is a mechanism to enforce problem-first thinking. Once the muscle exists, the same thinking can live in any format. Fetishising the artifact while skipping the discipline gets you the overhead with none of the benefit.
Running it on everything
There is a scale below which the overhead outweighs the value. Small features don't need a PR/FAQ; new products and significant bets do.
Is it for you?
Best for
Product leaders introducing rigour to a team that jumps straight from 'we have this capability' to 'let's ship it'
Not ideal for
Small incremental features, bug fixes, or fast-iteration environments where the overhead of the doc exceeds the value of the discipline
From the transcript
“you're adding the problem after the solution you're kind of retrofitting the problem retrofitting the customer”
“maybe if you don't have a problem paragraph there's not really a problem”
“solution paragraph and then the customer quote and then the fact which is like is there a legitimate plan to succeed”
“the key is that you don't just you don't think what we could build you think about the”
From the episode
What it takes to become a top 1% PM
Ian McAllister (Uber, Amazon, Airbnb)