What the double diamond asks a team to learn before it builds

The build estimate is already in the pack. The problem is a sentence the sponsor wrote on the way in. Nobody has sat with the person who meets that problem on a Tuesday.

The Design Council describes the Double Diamond as a picture of the design and innovation process. It is a simple account of the steps in a design and innovation project, whichever methods and tools a team happens to use. The Council licences the picture under CC BY 4.0. Richard Eisermann, then its Director of Design and Innovation, said the team codified the process, renamed the steps and popularised them. The four names on the picture are Discover, Define, Develop and Deliver.

Human-centred design, read off that picture, starts with the people who meet the problem. Corporate innovation starts to go wrong when a bet arrives with a build estimate and a sponsor's sentence, and the four names are treated as a slide title. The names are a sequence. A team can say them and still skip the part where someone sits with the work.

Discover, then define

Discover is the first diamond. The Council says it helps people understand the problem by speaking to the people affected and spending time with them, which is the alternative to assuming the problem. The person who meets it has to be in reach. A survey the sponsor already believes, or a persona card standing in for the operator, leaves that time unspent. The sentence in the pack should wait until the time has been spent.

I would write down what that person does, says and gets stuck on, in their words, before anyone offers a fix. The stuck point is the useful part. Official steps and the Tuesday workaround are often different, and the person doing the work is the one who can show you the difference. If they will only describe the official steps, you have learned something about the room, and you have not yet learned the work.

Define comes from what Discover produced. The Council says the insight gathered in discovery can help you define the challenge in a different way. The point of the phase is a challenge the people you spoke to would recognise. A sponsor's sentence that survives those conversations unchanged is a finding. If it was already right, you can show the observations that support it. If the conversations never had a chance to move it, the sentence was locked in the taxi. I want the changed sentence on the page, with the conversation that changed it written beside it.

Defining the challenge differently is the moment corporate innovation programmes usually lose their nerve. The original sentence has a sponsor, a steering date and a place in the portfolio, so a restated challenge looks like delay. That restatement is the work the first diamond was for. A portfolio should be able to show a before and an after for the problem, alongside whatever it shows for the solution. A page with only the original sentence is still a briefing.

Develop, then deliver

Develop is the second diamond. The Council says it encourages people to give different answers to the clearly defined problem, to look elsewhere for inspiration and to co-design with a range of people. The problem has to be clear before those answers are worth collecting. A wide pass at ideas, aimed at the sponsor's original sentence, produces options for a challenge you may not have. Quantity is useful once the challenge is one the affected person would recognise. Before that, quantity is a way of looking busy.

Co-design means the people affected are in the making, as well as in the research notes. A range of them, the Council says. One friendly user who likes the team is a thin version of that. The operator, a customer who had a bad Tuesday and someone who will have to change a handoff should all be able to mark up the same rough thing. When the only people in the room will present the concept, the second diamond has still to start.

Deliver, in the Council's words, is testing different solutions at small scale, rejecting the ones that will not work and improving the ones that will. The rejection is part of the phase. A portfolio review that scores every concept as continue, and never records a rejection, is still a show of the options. Small scale is the condition that makes a rejection affordable. A test that already has a build team, a launch date and a press line is large enough that a weak concept survives into the thing you then have to defend.

The order can loop. The Council notes that organisations it supports often learn more about the underlying problem, and that learning sends them back to the beginning of the diamond. Making and testing a very early idea can be part of discovery. In a digital setting, the Council says, no idea is ever finished. A corporate calendar that runs Discover on one date and Deliver on a later date, with no path back, has frozen the picture into a straight line. The picture itself allows a return.

What a portfolio owes the first diamond

I would put one line on every live bet, ahead of the build estimate. Who was spoken to, how much time the team spent with them and what in the challenge changed because of it. A blank line means Develop is still ideas about an assumption. Parking the bet is a legitimate outcome. So is a smaller discovery, with the operator in the room, before anyone prices a build.

The same line tells you whether human-centred design is operating or only named. If the notes describe the official process, and the operator's workaround is missing, the team observed the diagram. The next session should go back to the person and the stuck point, and the sponsor should stay long enough to hear it. A sponsor who leaves before that conversation trains the room to keep the tidy version.

Corporate innovation often keeps a second failure mode that looks like speed. The team accepts the sponsor's sentence, generates a wall of options and funds the one the senior person prefers. That copies Develop, leaves Discover and Define undone, and swaps Deliver's rejection step for a preference. The four names can still sit on the slide. The bet is a build aimed at a problem the team has yet to earn.

An AI tool can make that skip look like progress. A model can draft the options, the screen and the one-page concept on the afternoon the sentence arrives, so the pack looks further along. Until someone has spoken to the people affected, the draft is a fluent version of the assumption. I would keep the tool out of Develop until Define has a challenge those people would recognise. Clustering what they actually said is a fair use of the tool, and it still depends on the conversations having taken place.

Psychological safety sits under Discover. If describing a stuck point is expensive, the person gives you the official steps, and Define inherits the diagram you already had. A leader who wants the first diamond to be real makes the first honest description cheap, in front of the sponsor, before asking for ideas. Further interviews with people who feel unable to talk will keep returning that diagram.

What to run on one offer

1. Pick one live bet that already has a build estimate. Write the sponsor's current sentence at the top of a page.
2. Spend time with people who meet the problem, including the person who does the work when the official steps jam. Record what they do and where they get stuck, in their words.
3. Restate the challenge only from that record. If the sentence cannot change, show the observations that support it. If it can, change it before you collect options.
4. Co-design a small number of answers to that restated challenge, with a range of the people affected in the making.
5. Test those answers at a scale you can still reject. Record what you rejected, what you will improve and who owns the next version.
6. If the test teaches you that the challenge was still wrong, go back to Discover. The Council's own account is that this return is part of the work, and that an idea in a digital setting is not finished just because a phase has a date.

A design thinking workshop can hold that sequence for one offer, with the operator and the sponsor in the same week. The portfolio review after it still has to be able to show who was spoken to, what the challenge became and which options were rejected. A build estimate with none of those three is a bet that has not started the diamond.