“Consider organized anarchies.” (Cohen, March and Olsen, 1972)

Product management exists within the context of organized anarchies. Within that context, product managers do their best to create, read, update, and destroy elements of a product roadmap.

There are a few definitions of what a product roadmap is.

From Atlasssian: “A product roadmap is a shared source of truth that outlines the vision, direction, and progress of a product over time.”

Another: “A product roadmap can be seen as a key element of an organization that maps out the vision and direction of the product offering. It describes the way how a product or a product portfolio is going to meet a set of business objectives and the work that is required to get there.” (Münch, Trieflinger and Lang, 2019)

One of my favourite descriptions of a roadmap comes from Pablo Bouzada, below:

A misleading roadmap: “We will follow this linear path from zero to our inevitable victory, and these are the milestones that will get us there.

An honest roadmap: “We know where we are today, and have a good idea about the near future, but recognize that the future becomes uncertain.”

A strategic roadmap: “We know where we are today, and anticipate decision points that will change our path. We are intentionally gathering the data that will help us make those decisions.”

Certainty As A Good

Different people value certainty differently. Different organizations do too. There are some organizational cultures, what Laloux (2016) would call Blue Organizations with Blue Leadership that drive the uncertainty out of everything. Sometimes they do this to the point of valuing certainty over results.

Certainty wicks insecurity away from a decision.

There is a way to avoid the hazards of Blue leadership styles.

One heuristic comes from Colin Powell – the 40/70 rule. Put simply, if you make a decision with only 40% of the information in place, you’re calling it too early. If you want for more than 70% of the information, you’re calling it too late. The sweet-spot is between 40% and 70%.

False certainty, overconfidence, is a brutal mental bias. The definition of arrogance is undeserved confidence. The key word there is undeserved. Who’s doing the judging? Who decides? Even with a heuristic like the 40/70 rule, the number of known unknowns may explode over the course of a roadmap. Worse, sometimes, you are confronted with a decision opportunity for which there aren’t a whole lot similar situations to imitate. Novel problems sometimes take the courage to make novel decisions.

It’s against this context that a misleading roadmap is a desirable artifact for some organized anarchies, and, an absolutely toxic artifact in others.

The Vision Thing

Anybody can have a vision. An entrepreneur has a natural bias towards their vision.

Blank (2010, 2020) gives a recipe for a vision that’s pretty good. It goes like this: “For <customer segment> our product, <product name> is a <name the sector that customers say> that <benefit>. Unlike <competitors>, our product <discriminator>. Our product again is <product name>.”

There’s beauty in that formula because it lays bear the important factors. It’s unpopular because it lays bear the combinatorics involved in invention and there is an increased emphasis on story. Many of us growing up were heavily influenced by James Burke, so it doesn’t feel oppressive to some. To many entrepreneurs, the formula is oppression incarnate. It’s really just a recipe.

Typically, the founding entrepreneur comes up with a vision, and then rallies capital and resources to validate or reject that vision. Blank advises that this is a full contact sport, and, he sets out a pragmatic way to check for bias and gain certainty that there is a wonderful business worth building.

I’ll make a comparison to making a TV show or a movie. If you want to make changes to the product, you make them during the writing process. The costs of making changes during writing are non-zero, but relatively small compared to making changes in production. The costs soar when you’re making changes in post-production. They become ludicrous. Intuitively this makes sense. You’re changing around enormous volumes of data to come out with a different product.

The same is true of at the start of digital product. In the beginning, typically, a product is just words. If you want to make changes, it’s far cheaper to change those words early, during customer discovery and customer validation. Once you move into scale, hiring salespeople and all the infrastructure required to make something scale, the costs start to become ludicrous. Organizations begin to experience lock-in. The tension between certainty and the reversibility of decisions increases as facts accumulate and time passes.

It’s why Blank advises that you spend time objectively validating the vision for truth before you scale a business around it. It’s why there is such an emphasis on the term minimum and viable when referring to a minimum viable product.

There is considerable hazard in all of this. If a founder ignores too much of the evidence that their vision is invalid (either too soon, too late, or never viable), then they’ll have committed their firm to death. If a founder is too fast to change their mind, then they may never gather enough evidence for any vision at all. Colin Powell’s 40/70 rule is very useful in this context.

In this way, they can prematurely lock-in failure. Set sail for fail.

Or, far more unlikely, they can prematurely lock-in success.

Strategy Is Choice Amongst Weighted Alternatives

Too often, lists of desired features passes as a strategy. They are not a strategy because a list of things does not contain weightings.

I’m going to argue that an imitative list is not a strategy because there are no weighted alternatives.

A vision may contain a lot of clues that inform a strategy. Some of the best strategies are aligned against a vision. The choice to talk to a specific market, a group of people who refer to each other when making a purchasing decision (Moore, 1991). But a vision itself is not a strategy.

How are choices made? How are alternatives discovered? How are weights estimated?

The Garbage Can Model can be useful here.

Alternative generation is the easiest part of it all. Everybody has ideas. Product managers who possess facilitation skills can create spaces for large groups of heterogenous people to get together and share their ideas. Many facilitators like to create space for everybody by starting with quiet brainstorming. This is why so many conference rooms get blanketed in 3M stickies. One thing you should be aware of if you aren’t already: just because you’re an executive and you write something on a sticky, it doesn’t mean that it’s going to get executed. You aren’t generating a brief using a sticky.

Blank is careful to repeat that facts don’t exist within the organization – facts exist only with customers. There are typically severe risks for those who choose to refuse to canvas the organization for wisdom. There’s value in what people think that customers want. Sometimes statements about customer demand are really what people are wishing that customers are saying what they want. “Why can’t the focus group just say what we want them to say?”

In the Garbage Can Model, all problems are solutions looking for problems, and all solutions are problems looking for solutions. There’s really no differentiator between a problem and a solution.

This is a useful way of inviting ideas into the can so everybody can see them.

Often, customers present problems looking for solutions, or solutions looking for problems.

How are weights discovered?

Take another look at the Strategic Roadmap design above.

It looks a lot like a backcast canvas doesn’t it?

If you were to take that image and rotate it 90 degrees, you’d get a structure that looks like a decision tree.

Backcasting and decision trees have so much in common that it’s difficult to tell them apart. Typically, when you’re constructing a decision tree, you’re imagining outcomes and then assigning values, or weights, to each of the leaves. Then, based on the weights along the leaves of the tree, you work your way up the tree, deleting branches from consideration. You’re often left with a single leaf down at the bottom and a course of desirable action to take. (Extra spicy Bayesian friends may also combine weights with probabilities and run through the same exercise).

Simplified backcasting involves being honest with where you’re starting from, then imagining a worst, best, and median future and building the decision tree from that material. While anybody can have a vision, it takes a particular mindset to engage in backcasting. Product managers can develop awareness of which personalities actively block themselves, and others, from engaging in discussions about the future.

Standard management science advises that you normalize all the outcomes using a currency. Blue leadership styles will insist on calculating a Return On Investment before entering the quagmire of asking over which time period to amortize the benefits. Teal leadership emphasizes a common purpose.

How Are Choices Made?

Because a key feature of Garbage Cans is variable energy and fluid participation, the nature of decision of making is often stochastic.

Most Garbage Cans are dominated by internal actors within the organization. Some might include prominent customers. Some might accurately or inaccurately proxy their customers. Some represent their departments interests. Some represent the firms interests. Some proxy. Some don’t. There’s a simplifying way through.

Facts don’t exist inside the building (Blank, 2010, 2020).

By asking yourself which assumptions need to be true at different decision points, you can identify hypotheses and related experiments that have to happen, in which sequence, to arrive at a desirable point in the future. This offers a kind of smooth rationality that is lacking in a pure Garbage Can. It makes for guided fact generation from customers themselves, rather than relying on the preference structures of those engaged in the Garbage Can. Moreover, such a flow aligns well with the rational customer discovery and customer validation flows found in Blank (2010, 2020).

The technique at least offers those who value certainty over results to be able to understand the set of facts that must be true before an action they certainly want to be taken will be taken. It does not offer them the certainty that an action will be taken by a specific day and time.

In this way, decisions opportunities are identified in advance, and evidence is gathered deliberately at each step to inform the next decision opportunity. Desired futures may change as uncertainty is resolved or as the organizations appetite for risk changes over time. This is to say, the roadmap isn’t a locked-in set of promises, but rather, a real map of potential roads to take to arrive at different destinations.

Moreover, decision by flight is less likely to occur when assumptions are laid bare and the roadmap is about deliberately reducing the uncertainty on specific lines of decisions, while deliberately trying to attain some outcome that is aligned with the vision. It may even offer a scaffolding for self-directed OKR’s. This feels more participatory and Teal (Laloux, 2016), doesn’t it?

Choices, ideally, are made deliberately, within the strategic context they appear. When organizational factors dominate the Garbage Can, learning and progress halts.

Too much anarchy, and nobody learns a thing. The discussion never advances and very little progress can be made. If any is, it’s purely by chance.

Too little anarchy, too much certainty, and nobody learns a thing. The discussion never advances and very little progress can be made. If any is, it’s purely by chance.

Powell’s 40/70 Rule offers a window into what balanced chaos and a shared sense of curiosity, with just enough security to be viable, can offer.

A Garbage Can Product Roadmap

In this post, I’ve described a variant of Bouzada’s Strategic Roadmap with the harsh environment of the Garbage Can Model (1972), orthodox Blank (2010, 2020) and Laloux (2016).

There many enabling conditions to building a Garbage Can Product Roadmap, that:

  • There can be transparent discussions about the validity of a Vision.
  • Everybody has wisdom to contribute in generating alternatives.
  • A group of people can engage in a discussion about alternate futures.
  • Alternatives can be weighted.
  • Facts from outside the building will be trusted or factored.
  • Assumptions are recognized as assumptions.
  • Assumptions can be structured into a tree or tree-like sequence.
  • Assumptions can be tested.
  • Test results will be trusted.
  • The 40/70 rule provides enough security for uncertainty to be resolved.

It’s a long list

It’s probably in everybody’s best interest for product management to smooth the worst features of the Garbage Can while embracing some its best features.


Blank, S. (2020). The four steps to the epiphany: successful strategies for products that win. John Wiley & Sons.

Cohen, M. D., March, J. G., & Olsen, J. P. (1972). A garbage can model of organizational choice. Administrative science quarterly, 1-25.

Laloux, F. (2016). Reinventing organizations. LannooCampus.

Moore, G. A., & McKenna, R. (1991). Crossing the chasm.

Münch, J., Trieflinger, S., & Lang, D. (2019, June). Product roadmap–from vision to reality: a systematic literature review. In 2019 IEEE International Conference on Engineering, Technology and Innovation (ICE/ITMC) (pp. 1-8). IEEE.