(Photo by Alvaro Reyes on Unsplash)

Want to improve your AI-based products and services? Would you like to see a better return on your AI investment, while reducing failures and false starts? One simple idea can make a large difference:

Involve your data team in product planning efforts.

And I mean, early.

I emphasize the word "involve" there. Don't just invite them to meetings and ask them to sit quietly. They should be active participants. You need to bring them up to speed on what you're planning to build and why. Encourage them to ask questions and offer their ideas. Ask them your questions and then – here's the important part – be prepared to listen to their answers.

When I share this idea with executives and product owners, I get one of three reactions:

1. "My company already does this. I thought everyone did this. Who doesn't do this??"
2. "I know this is a good idea and I'd like to do it, but I'd get stiff resistance."
3. "What? Why the hell would I do that?"

And these are just the verbal answers. I recall one person who simply scoffed.

(Who still scoffs, by the way? What year is this, 1725?)

For people in the first group: Congrats! You're in a good place. You're also in a rare space.

If you're in the second group: There's still hope. Stick around for ideas on how to approach this.

For the third group: What I've said runs contrary to your beliefs, so you're tempted to close this browser tab. Feel free! But to reiterate my opening point: what I have to say could save you time, money, effort, and reputation on your company's AI efforts.

Why should you involve the data team?

Your company's data practitioners – the data scientists, data engineers, AI engineers, and any other title(s) you use – are your local experts on all things data science, ML, and AI. They don't just do data work; they* understand how data works.* They know what these technologies can do, what they cannot do, and most importantly for a product, they know when these tools appear to be in working order but are actually on the brink of an incident.

If you're building any product that embeds or relies on data, you want that level of expertise in the room early and often. This is how you catch potential problems long before the product rollout – sparing you from having to walk back promises you've made to clients, sales prospects, stakeholders, and even regulators. Your company's data professionals can also note when your idea doesn't reach far enough, because it doesn't take advantage of what AI has to offer.

It's possible that your product team and stakeholders have already developed the requisite AI literacy; but in my experience, that's rare. It's far more likely that there are gaps in your knowledge of AI. (Which is understandable, as AI is not your field of practice!) Those knowledge gaps can lead you to adopt an idea that's out of reach – maybe due to small issues, maybe because of a show-stopper, maybe because of corner cases you hadn't considered.

Excuses, excuses

If you want to involve the data team in early product discussions, but face resistance from your peers, there's still hope. Here I've provided some weak excuses reasons you've probably heard before, along with suitable responses:

• "Oh, the data team is too busy." If you think they're busy now, wait until they have to scramble to build something you've promised!
• "We'll loop them in after we've decided what to build." Wouldn't you want them to confirm that your use of AI is viable before the proverbial cement hardens?
• "The data team doesn't understand our business well enough." Maybe, maybe not. Have you considered bringing them up to speed on the company's business model? They'll provide a ton more value if they have that knowledge.
• "I'm the product manager so I get to decide what to build." You are ultimately responsible for what the company builds, yes. Given that, wouldn't you want expert eyes on your plans before you've promised the world what's coming?

There are plenty more excuses, I'm sure. You can counter most of them with:

We have AI expertise on-hand, and you're refusing to consult with them on AI matters? Really?

Better in the long run

Still not convinced? I acknowledge that seeking input from the data team – or anyone else, for that matter – can slow down a process that probably requires too many sign-offs as-is.

That said, remember that the data team's goal is not to slow you down. They, like you, want to build the best products and make the company successful. Involving them early and often on any AI-related product matters can help avoid false starts, reduce long-term costs, identify downside risk exposures, and uncover missed opportunities.