The data lake trap:
Why many AI strategies stall

Amir Ouki Autonomous Innovation Summit

Amir Ouki

Managing Director,

Applied AI & Technology

Many companies kick off their AI journey by investing in large-scale data lake initiatives: terabytes of historical data are centralized, cleaned, tagged, and stored with the hope that it will serve as a launchpad for AI.

The assumption is simple: to do AI right, you need all your data in one place first.

But in reality, that’s often the wrong sequence.

Data lakes aren’t always a prerequisite for AI. They’re an outcome.

The most effective AI initiatives begin not with storage but with clear business goals. To support that, you need the right data, not all the data.

AI is a capability for improving decisions:

  • What to recommend?
  • When to intervene?
  • How to optimize an outcome?

A focused use case (like reducing cart abandonment or predicting maintenance failures) allows you to work backward to identify the minimum viable dataset required.

Once you’ve proven that the multiple AI solutions deliver impact, you can invest in infrastructure that makes them repeatable and scalable. That might include a data lake, but now it’s in service of something, not speculative.

Building for storage vs. building for decisions

The temptation to centralize all data is understandable. It feels like a future-proofing move. But without clear use cases driving it, a data lake often becomes a data swamp: difficult to navigate, hard to trust, and expensive to maintain.

The alternative is to build just enough.

When the starting point is a valuable decision worth improving, the data stack can be smaller, faster, and far more targeted. Instead of a massive upfront investment, you build incrementally as your needs evolve.

What comes first: the AI or the lake?

Neither. What comes first is the decision.

AI strategy should begin with identifying high-impact decisions that could benefit from augmentation or automation. Once you know the decision, you can figure out the data needed to support it. Then you can define how to collect, store, and access that data.

This decision-first sequencing is far more effective than a bottom-up infrastructure-led approach, which risks creating technical complexity without strategic clarity.

When a data lake is the right first step

There are clearly situations where starting with a data lake is justified. If your organization already has a well-defined set of AI or analytics use cases across departments and you’re facing systemic issues with data access, quality, or duplication, a data lake can help create coherence.

In these cases, a lake can create leverage. But even then, its structure should be defined by the actual decisions and use cases it needs to support. Not by the urge to centralize everything just in case.

AI strategy ≠ data lake strategy

AI success is about shortening the loop between insight and action. You don’t need a decade of historical data to improve how decisions are made tomorrow. You need usable, relevant data tied to specific business outcomes.

Before you greenlight your next big data lake initiative, ask yourself: what decisions are we trying to improve, and what is the fastest path to improving them?

You may find the lake can wait.

AI Strategy Playbook

This AI Strategy Playbook is your guide to win in the age of AI. Learn how to build AI capabilities around real business decisions.

Managing Director, Applied AI & Technology

Amir leads BOI’s global team of product strategists, designers, and engineers in designing and building AI technology that transforms roles, functions, and businesses. Amir loves to solve complex real world challenges that have an immediate impact, and is especially focused on KPI-led software that drives growth and innovation across the top and bottom line. He can often be found (objectively) evaluating and assessing new technologies that could benefit our clients and has launched products with Anthropic, Apple, Netflix, Palantir, Google, Twitch, Bank of America, and others.

Keep learning

More resources on AI Strategy