Data-Informed, Intuition-Led: Decoding B2B-focused CV-enabled Product Management

Data-based decisions are nice. It’s safe. Data from app-based user analytics and user behavior observations driving Product decisions is a surefire way to improve product adoption and user retention.

But getting Data based observations esp. as a startup is difficult esp. if your product is B2B focused. There is a high chance that your first five customers may not even be your ideal customer profile. The user data when collected is too limited and diverse to make any meaningful derivations.

This problem is only compounded when as a startup, you do founder-led sales and get an enterprise account as your 1st customer. With every product decision dictated more by the Enterprise account’s security and legal team along with the end-users who more often than not require changes based on their specific use cases the chance to build a product that can be scalable across the mid-market and SMBs is lost.

Now add computer vision into the mix. With long development cycles, the ability to make the right product decisions early in the data collection cycle is crucial. The most important decisions would mostly be around specificity i.e deciding what is the specific set of items your CV algorithms will identify based on the use case and which it will neglect. This makes or breaks the product.

So, early in the game, you have to make decisions on what your algo should do and most importantly shouldn’t do even before you have a chance to get your first paying customer. Many a time, the solution you set out to build will be the first of its kind for that specific problem as no one would have attempted to use computer vision to solve that particular issue. In that scenario, you not only have to define the product specs, but you also have to hypothesize on potential user behavior.

In this scenario, while data is a good thing to have most of the time we are Intuition-led. THat’s especially the case when building B2B focussed Computer Vision enabled products.

But then the question arises, how do we manage this Intuition flow and make sure we make the right decisions both by ourselves and as a team.

Let’s be clear; data is of paramount importance; it’s the bedrock on which realistic and viable product decisions are made. Quantitative evidence is vital in making informed decisions that shape the course of product development.

However, the emphasis is on the term “informed.” Data informs decisions; it doesn’t dictate them.

Intuition is not a disregard for data; it’s rather an advanced interpretation of it. Good Intuition is backed up by positive data sooner rather than later.

So, how to bring in Intution?

The first and most obv step is to listen to the customer’s problem but not to listen to what the customer thinks is the solution. Having this means that we can list their issues without being encumbered by their solution ideas.

The reason why this is important is that esp. with computer vision based solutions it’s not going to be incremental improvement of current process but rather complete automation of the part of the process eg. would be reducing the warehouse associate touch points required to ship a parcel from 20 to 4-5.

So that’s one. The next is all about cultivating a workplace culture that values both data-enabled insights and intutive optinions. It’s really difficult to have an engineering team esp. those in R&D to share ideas to management. This is even more difficult if the opinion they have run counter to the data collected till date.

I have found that in such a meeting, to get everyone to open up, we need to start with data. Talk about the data collected esp. data everone can agree on and then slowly get into what I believe to be contentious data and start adding my opinions. This way people start getting comfortable on sharing ideas that might even be wrong.

Encouraging such a culture means recognizing and rewarding not just successful outcomes but also willingness to take risks. It also means creating an environment where ‘failing fast’ is seen as a virtue.

But that means we need to have a team that’s just not strong technically but also must have strong extrapolation skills.

Extrapolation skills here refer not just to the ability to project data trends into the future but from a tech perspective anticipate issues that can arise as we scale and from a product perspective envision user behavior changes and pitfalls i.e which behaviors can change and which will not in the absence of concrete data

But building a team that can use the existing data without getting bogged down whilst also having the ability to connect disparate dots, and identify potential disruption not just in scalability but in user behavior is difficult.

Recruitment here involves having a keen eye for talent that exhibits technical prowess and creative thinking. Does the individual demonstrate a curiosity beyond their field of expertise in addition to being comfortable with ambiguity?

I could go on. But recruiting and retaining the right team, and providing the right environment go a long way in getting that half-baked thought processes out into the open for discussion.

But what happens when data clashes with intuition?

I wish I had a straightforward answer for this one. In my case, I just keep talking until I get some idea. Depending on the team culture, stage of the company etc maybe there can be some frameworks that can provide some empirical basis for decision making.

But how can I get into a framework of a gut instinct that draws upon an often subconscious understanding of the market and technology, drawing on a well of experience, expertise, and what not.

How can I create a framework that combines the above needs but also balances the risks of biases or overly optimistic predictions ( This is a huge problem in itself)

Frank;y, I do not have an answer to this currently. Maybe, in the future. But for now, the plan is to keep talking until an idea pops up.

Simply put, the interplay between data and intuition is incredibly useful in B2B CV product management.

How a Product Manager is going to manage this will be something interesting to observe.