I led over 100 projects in 4 years as a data consultant.

The truth is that 95% of these projects did not involve machine learning. And that's a good thing. Most companies should take an "ML as a last resort" approach to extracting value from data.

We put that mindset into practice to double revenue (and win ML work in the process).

Land-and-expand your data consulting projects

In other words, start small.

Form a strong relationship with the company through exceptional service.

Upsell across different projects and expand your services over time.

And here's what that looked like for our team.

#1: Organize the data

Most companies need significant engineering help.

• Write stored procedures to normalize data

• Install best practices among sales, marketing, and finance teams

The keys in this phase: simplicity and durability.

#2: Analyze the data

Unlock the first level of value by producing consistent analytics to measure performance.

• Define a set of key business analytics

• Design a reporting workflow and product

The keys in this phase: reusability and insight.

#3: Model the data

Use predictive modeling to identify and action on trends before they happen.

• Build models to target leads, reduce churn, and identify upsell opportunity

• Deliver actionable insights to stakeholders

They keys in this phase: actionability and value.