Dear Advisor: I’d like to makE data-driven decisions. Where do I start?

Post was originally published in May 2021 and has been updated in July 2021 for relevancy.

Great question! The short answer is that it depends… The path to becoming data-driven is non-linear, because it depends on many things, including:

  • what the executive priorities for data-driven projects are,

  • knowing what is -- and isn’t -- getting collected,

  • what the skills and bandwidth of the team(s) are.

Having said that, here are some guidelines to get started.

First, let’s take a step back and understand the motivation for this question:

  • Who and what is the driving force behind this?

  • Is there executive buy-in?

  • Why now?

  • What is the ultimate goal that data may help the company accomplish?

Next, discuss and identify the most pressing question(s) the executive team would like to answer that would also bring value to the business in the short-term -- and decide on 1 question to try to answer as proof of concept (POC).

  • Here's my framework to help you prioritize, scope down and learn from.

  • Note: Starting a POC that aims to answer a key business question in a relatively short amount of time will: build rapport across the company in the short term and help you get support for making investments into data in the long term (vs starting with the end goal of documenting data architecture or an ML/AI research question where business impact is less clear).

  • Sample topics to explore for POC:

      • What does repeat engagement look like for your B2B/B2C customers?

      • Do your executives know the answers to their most pressing question(s)?

      • What will it take to get your key (client) report scalable?

As you execute on the POC to address above, you’ll get an additional finding -- a (partial) data and strategy audit:

  • evaluating if data is available to answer the question,

  • recognize what’s hard to access/reproduce,

  • identify what's not getting collected,

  • get an idea of what the (incomplete) data architecture looks like, and

  • if there are any other constraints/considerations.

These findings can help you build further rapport with stakeholders. They'll also help you get everyone on the same page by adding any findings of gaps in product/data strategy to the product backlog for prioritization and iteration as a company.

What’s out-of-scope for POC:

  • Guarantee that you’ll be able to answer the above question, since data may not available.

    • Recommendation: Check-in with stakeholders as soon as possible if there are blockers to answering the question -- and provide recommendations for next steps.

  • Guarantee that it will take a predetermined number of hours to accomplish.

    • Recommendation: Time-box this step to help you scope down and scope out the ask.

  • Making it production-ready, scalable with low latency, etc. Once you have a highly visible POC, the data audit will help you answer what's needed to get it in real-time and/or scale it.

What if there isn't executive buy-in? What can you do?

I still recommend starting out with a very scoped-down, time-boxed, POC to try to answer a key business question, potentially as a 20% project. The results of the POC will help you do 1 of 2 things:

    • start the conversation around the value of the data, or

    • (on the flip side) help you see that while key pieces are missing there is potential to answer other business questions.

Do you need an expert to help you identify the most immediate business needs and/or help you execute a POC? Please reach out.

Where to go from here?