Where to Focus

Data minimalism means reducing the number of disparate analytical and AI initiatives in favor of the few bets that are likely to pay off. In 2025, most companies should be focusing data spend on specific, strategic areas and avoiding ad-hoc and highly experimental developments. Many data teams go off into left field without having first built the floor plan -- the schema and metrics that the entire business revolves around. Hakuin is expert at driving foundational capabilities. We focus on and deliver the basics first.

The purpose of data and AI is to improve decision quality - either the intelligence level of the decision or the speed at which the decision can be executed. To achieve this, work backwards from the available actions and decisions.

Hakuin focuses on the Inquiry Framework to cull those initiatives unlikely to deliver value in favor of those with a clear path to increasing enterprise value. We have the experience and expertise to deliver value while avoiding the common pitfalls, emphasizing centralization and clarity first.

Get in touch

Outputs from Data

Data outputs come in only a few forms. The first form is analytical output. This is most commonly a visual reporting format, either shared as a dashboard, notebook, or presentation. The purpose of analytical outputs is to improve human decisioning and awareness. It is decision support. The first step is to determine the range of topics and decisions requiring support and develop the appropriate structures, pipelines, and delivery formats. If this assembly line is developed piecemeal, it can have costly implications downstream.

The key here is to reduce the variety and range of outputs and to centralize to the greatest extent possible. Metrics are the building blocks of business intelligence, and we need a set of metrics that relate to each other. If we build hierarchies of metrics or team specific metrics, we lose the connective tissue that enables us to understand how metrics relate to one another (which is the purpose of having metrics in the first place).

ML & AI

ML & AI development is also intended for decision improvement. Most ML & AI developments are intended to make decisions on behalf of humans- for example, the recommendation models that run on e-commerce sites make product suggestions to users without the intermediation of human decision makers. In this case, ML & AI are meant to scale decisions.

Starting Simple

We have to always ask ourselves whether a given AI model could be accomplished with fewer resources or at least started off with something more simple - for example, conditional rules. At the very least, we should always build the most simple version first. This often means conditional rules based models. The conditional rules based model sets a benchmark for performance and also improves our understanding of the problem space before applying ML and AI algorithms.

Lifetime Value

‍

Data and AI have the power to transform business operations. But the path towards success requires careful scrutiny and skepticism.

Hakuin has the expertise and rich experience to identify the rich opportunities for companies to improve efficiency and decision intelligence, without falling into the common traps.

‍