Below are ten 2022 predictions for data product leaders and organizations trying to leverage ML and analytics in their software, tools, apps, and services.

From the lens of a consulting product designer.

Yea, you head that one right. If that PhD in physics, statistics, math or engineering in you is already making you cringe, you may depart this article through gate 12—I won't be mad!

Why do I talk about this?

There are enough curious, aspiring, and experienced data product leaders out there who know they need to focus a lot more on the people, outcomes, and experience side of technology that uses ML/AI and traditional analytics. They've built enough technically right, but effectively wrong data products to know that data science, analytics, statistics, and machine learning engineering isn't enough to make people care, to adopt decision-support technology, and to trust AI.

My predictions this year are entirely based on qualitative research, discussions, exposure to you, my clients, and people I talk to - no big data set behind it! If you want that, just go down 3 blocks, turn right past the Dunkin with the wicked ugly kid outside sellin' newspapahs, hop on the red line, change at downtown crossing, head through the old Filene's Basement exit, and you'll find an API end point. If you want the key to the API or the bathroom, you gotta go back and ask the ugly kid.

And now, this:

1. Data products that don't involve users and stakeholders heavily upstream in the research and design phase will continue to result in low-use/no-use outputs that do not produce outcomes. If you don't get access to the ACTUAL end-users of the product or service, you're likely going to fail - even if the technology effort seems to be progressing.
2. AI/Data Product Manager roles will continue to grow, but slowly. (Note: this was IIA's #4 prediction for this year). I think most of the people in these roles will not be from digital PMs for a variety of reasons, but good companies will try to attract that type of talent.
3. There will be slow, but some growth in the area of "product" vs. project based approaches to building data products in traditional enterprise companies. (In part due to #2).
4. Traditional analytics and data science teams at traditional enterprise orgs. will begin to see where a lack of a designed or defined UX is hurting their work, and there may be some hiring (perhaps even less than #2).
5. Consulting firms in data science and analytics that being to bring design resources on board will likely deliver better data products and services to clients and see this as a competitive differentiator after first going through a "hey, this was helpful to us internally" phase first.
6. In the UX and design field, there will be more sub-specialization groups and discussions emerging (adding on to ones like Michelle Carney's MLUX, Ben Shneiderman's HCAI book launch + group, Nadia Piet's AI x Design, etc.) However, the overall numbers will remain low, and the discussions here will either skew heavily into the ethics realm, or will be biased by the largest digital-natives/FANG-type companies' perspectives and work.
7. Design leaders, and to a lesser extent, product leaders at established software companies will continue to have their head in the sand (sorry!) about their roles and the opportunities around data and ML/AI. Outside the more mature FANG companies etc, AI will mostly continue to mean "chatbots" and recommender algos -- "features" that don't appear to really require design thinking or input. Digital/design teams in traditional enterprise orgs. will not have any formal or centralized perspective about data yet, nor how it can improve the UX customers, employees, partners, vendors, etc.
8. Orgs that hire lead data scientists that have (or are willing to develop) good interpersonal, creative, and non-technical skills will see outsized advantage, particularly as ML initiatives rely less on hand-coded, hand-built solutions, and will face their greatest challenges in the adoption/usability/trust phase of the projects. (aka "operationalization" but we don't use that word 'round these parts!)
9. "Problem-owning teams" will see larger successes vs. "IT/data teams" who are "assigned" to business projects—particularly in the large traditional enterprises who are using ML/AI inside of custom applications that require users to "change" their behavior. The more the org's see data science/digital/analytics as partners and part of the business teams, the more value will be created.
10. I will be lucky to get 3/10 of these right for 2022, but nobody will know because they didn't listen to episode 80 of my podcast with Doug Hubbard on how to measure anything (but especially value in data products!).
11. Fun fact: ML won't get renamed "machine guessing" as much as I love that framing.

That's it for the predictions, but if you want to learn more about where all this is coming from, subscribe to my Insights mailing list below. I regularly write and podcast about how data product leaders can, should, and are using design-driven innovation to create indispensable ML and analytics solutions.

Ready?

2022 is here; let's go.

Photo by Sharon McCutcheon on Unsplash