What the first small bets taught

The point of making small bets in the first place

There is a version of building that runs on conviction alone, where the plan is clear, the model is tight, and the only remaining task is execution. It is a compelling way to feel about your work. It is also, in most cases, wrong.

Small bets are not hedging. They are not a lack of commitment dressed up in reasonable-sounding language. They are how you learn something true faster than thinking would allow. The thinking comes before and after, but only the doing produces the kind of information worth reasoning from. That is the underlying premise of everything attempted this year, and it is worth naming clearly before getting into what happened.

Mustard Seed Group in 2015 was a holding structure in the earliest stages of building. No product was mature. No revenue was structural. No team was large. The year was about finding which directions were real: which things, when tried, produced the kind of friction and engagement and interest that suggests there is something actually there.

Here is what we learned.

---

What we were wrong about first

The first incorrect assumption was about timelines. There is a version of optimism that convinces you things will move faster than they do, not because you are naïve about technology, but because you are naïve about the gap between something working in a controlled environment and something being ready to be relied upon. That gap is almost always larger than expected and almost always more instructive than you anticipated.

The second incorrect assumption was about which parts of the work would be hard. Early on, it seemed like the hard part would be making things function. Getting data to flow, getting interfaces to render, getting logic to execute reliably. These were the problems that occupied the most mental space in planning. As it turned out, they were the tractable parts. The harder parts were the ones that resisted pure engineering: what does this need to feel like? Who is it actually for? What do they need that they have not articulated? What should the product refuse to do?

These are not questions you answer through thinking. You answer them by building something real enough to react to, then watching closely.

The third incorrect assumption was subtler. There was an early instinct that each project should be kept cleanly separate, such that mixing concerns would muddy the output. That turned out to be partially wrong. The cleanest insights of the year came from noticing how patterns observed in one context were relevant to problems in another. The consumer and the commercial, the research and the product, the services work and the platform ambition: these were not separate worlds. They were, when observed together, the same underlying question about capability, expressed differently.

---

What the experiments actually produced

None of the early experiments produced finished products. That was not the goal. What they produced was a much cleaner picture of the problem space, a more honest understanding of technical constraints, and a set of working hypotheses that were grounded in evidence rather than inference.

The most valuable output was negative space: a clearer understanding of what was not the right direction. This sounds disappointing if you measure progress by what gets built. It is not disappointing at all if you understand that eliminating wrong directions early is far cheaper than pursuing them to completion.

There was also a clearer sense of what made the work worthwhile in the first place. Not the opportunity, not the market size, not the projected trajectory. The specific capability gap that felt genuinely important to close. The concrete thing that should exist but does not. That clarity is not a small thing. Most building fails because the thing being built was never well-defined enough to aim at. Getting the definition right is itself a substantial piece of work.

What took shape across 2015 was the recognisable outline of the portfolio that Mustard Seed Group is becoming. The division between research, products, services, and consumer experiences: that structure did not arrive by design from the first day. It emerged from watching which types of work naturally pulled in which directions, which problems wanted to be solved by a long-term platform and which ones wanted to be solved by a fast commercial engagement, which capabilities needed to be owned internally and which could be acquired or activated through partners.

---

The thing about shipping imperfect things

There is significant resistance, both personal and cultural, to releasing work that is not yet ready. The definition of "ready" is the problem. If ready means polished and complete and capable of sustaining scrutiny from a demanding audience, then almost nothing is ever ready. If ready means real enough to generate a genuine reaction, then it becomes possible to move.

The discipline being built here is the second kind. Not recklessness. Not indifference to quality. But a working acceptance that the feedback you get from something imperfect and real is worth more than the confidence you derive from something perfect and hypothetical. Reality has information that planning does not.

This is difficult to internalise because the imperfect version reflects on you. The finished version does not exist yet, so it can be as good as the imagination will allow. The shipped version is visible and flawed and committed to. But that discomfort is precisely the mechanism that makes it useful. You cannot learn from something you are still imagining.

The lesson from 2015 is not that quality does not matter. It is that there is a specific order of operations. You get to the quality by shipping the imperfect version first, not by withholding until the perfect version is ready. The imperfect version tells you whether you are aiming at the right thing before you invest the effort required to do it properly.

---

What 2016 should look like from here

The lessons above have a logical implication for the year ahead, and it is not a gentle one. If the right move in 2015 was to make small bets and observe what they taught, the right move in 2016 is to take those lessons seriously enough to let them change the direction of the work, not just confirm that the original plan was correct.

This means several things practically.

The research layer needs to become more systematic. The insights from early experiments were valuable but uneven. Some things were observed carefully and recorded usefully. Others were felt rather than documented, which means they are harder to act on and easier to misremember. In 2016, the goal is to build better habits around capturing what is actually learned, not in a bureaucratic sense, but in the sense that learning that is not captured is learning that does not compound.

The services dimension needs to become a deliberate testing environment. When you work directly with clients on real problems using your own tools and methods, you get something that product development in isolation cannot give you: the experience of your approach meeting someone else's reality at high stakes. That contact is irreplaceable. It sharpens the product intuition faster than any other mechanism.

The consumer work needs to stay honest about what it is still figuring out. There is more to understand about the specific people it is for and what it needs to do for them that they cannot currently get elsewhere. That question needs to keep being asked with genuine openness rather than answered prematurely to make forward motion feel more assured than it is.

And the portfolio as a whole needs to keep being organised around the same centre: not a technology, not a market, but a question. What makes people more capable? What expands what they can think, what they can create, what they can execute? That question is not going to have a single answer. The point of having a portfolio is that it can hold multiple bets against the same thesis at the same time.

---

One clear thing from this year

If 2015 taught one thing above everything else, it is this: the value of an idea is not in the idea. It is in what survives contact with the work of making it real.

A large number of things that seemed compelling in January of this year were revealed, by the time of doing them, to be less interesting than they appeared. And a smaller number of things that seemed like maintenance tasks or secondary concerns turned out to be load-bearing: the places where the real work was hiding. You cannot know which is which in advance. You can only run the experiment.

That is what 2015 was. It was the running of experiments whose results we are now in a position to read honestly.

The work continues.