(Image credit: Jason Goodman on Unsplash)

Data science practitioners are often told that they need to "learn how to communicate" with the people outside of their team … but they rarely get any direction other than "use less technical jargon." That's a start, but it's not really helpful.

Would you like three clear, concrete steps to improve your communication with stakeholders and product owners? Read on.

Step 1: Share your results

Now, you know that you've been working very hard on updating that ML model. But how would your stakeholders know that? ML work is a black box for many executives. Sometimes they can't tell the difference between you just sitting at your desk and you actually getting work done.

The solution? **Show them your progress. ** Keep a running log of what changes you've made to the model – new tuning parameters, new dataset, and the like – and post it in a place where stakeholders and product owners can see it.

You don't have to do anything fancy here. You only need four steps:

1. Create a shared spreadsheet. Call it "Model Scoreboard" or whatever.
2. Create two tabs, "Description" and "Reporting."
3. In the Reporting tab: define columns for the model name, date, the changes you tried, the metrics, and any additional notes. This will hold the information for each model run.
4. In the Description tab: describe the purpose of the spreadsheet and explain the fields you used in the Reporting tab. This will be especially helpful to people outside of the team who want to understand what's going on.

Now all you have to do is spread the word. Post a link to the spreadsheet in Slack, mention it in any meetings you hold with stakeholders, and ask your data science teammates to do the same. By making this accessible to everyone, you get the added bonus that your stakeholders will never have to ask you for it. They'll already know where to go.

You might question whether sharing a spreadsheet counts as "communication." Fair enough. Remember that communication is more than conversation. It is about conveying information. That's exactly what this spreadsheet does.

Keep in mind, this spreadsheet doesn't just show model results; it shows people that you're actually working on something. It helps decision-makers determine when to continue a given modeling effort versus when to pull the plug.

Step 2: Speak their language

Some execs have a nose for technology and actually want to hear the deep details of what you're doing. Great! If they ask for that information, you're welcome to provide it.

But for the most part, these people are short on time and need to make a decision based on a simplified picture of your findings. So be prepared to deliver just that.

Executives see a company's operations -- not just the AI portion, but the entire organization -- through four lenses:

• Where are we making more money?
• Where are we losing money?
• How are we spotting new opportunities?
• How are we managing risk?

Anything you bring to their doorstep – from analysis results to budget requests – you'll want to frame it in those terms.

But how do you do that? Well, that leads us to …

Step 3: Learn about the business

A lot of people in tech have asked me for career advice over the years. One consistent message I've delivered is to learn the company's business model. What do they actually do? What are the goals and challenges of this company, and of its wider industry?

"Fair enough," you say. "But how do I learn this?"

Why not start with your employer's website? This is what sales prospects and current customers see. The descriptions and marketing materials posted there should explain what services are provided and who would derive benefit. This is useful for helping you understand what the company actually does.

Next, schedule a quick chat with your manager. Ask them, flat-out:

• "How does this company make money, and how does data science (or ML, or AI, or whatever) play a role in that?"
• "What resources would you recommend I review so that I can learn more about this field?"

In most cases, your manager will be eager to help. Perhaps even overjoyed. (If they discourage you from learning that kind of information, well … consider what they're really saying.)

Why not try it out?

You'll notice: I haven't suggested that you sugar-coat details. Nor have I proposed heavyweight frameworks or commercial tools. These three tips are well within your reach.

Items 2 and 3 will take some time to develop but require only a little time to get started. And you can accomplish Item 1 in minutes. Combined, these will dramatically improve how you communicate with your stakeholders and product managers, which should smooth your relationship with them and make you more effective in your role.