Why Most Data Projects Fail & How to Avoid It • Jesse Anderson • GOTO 2023

Interesting. speech Jesse Anderson, author of Data Teams. The author discusses the key questions to ask when starting projects

• Who The author talks about the right team composition for data projects. Actually, the author has written a whole book about this and he talks about the balance of data scientists, data engineers, operations.
• What The author asks about the business value of the data product / project that you are engaged in. The CEO’s phrase “We’re doing AI” is missing from the data strategy.) Understand how your project will bring value to your business. In addition to strategy, you need a plan and execution. Especially at a time when tech companies are making cuts in ways that don’t make money.
• When The author talks about when this business value will be created. You need a project with clear time limits, so that it's not too long to be canceled somewhere in the middle and not too short, promising golden mountains that will actually be impossible to match.
• Where And now we finally got to the first technical question, and where exactly this data will be processed, what the architecture of the solution will look like. And here the answer is also missing the phrase "We will use the technology XYZ vendor ABC". The problem is that the vendor can promise anything, but this promise is not a fact that it is feasible, moreover, it is not a fact that it is optimal for the customer:)
• How We are talking about the implementation plan and focusing on priority areas. Although often such data projects try to do everything at once, and then lose their effectiveness on context switches and freeze in place, no longer generating any value other than stories about the onset of AI:) The author interestingly tells about how business customers are perpendicular to specific technical solutions, but it is important what business value they can get based on the results of the plan.
• Why The question is, why does this data have value? Simply shipping data and driving ETL/ELT Piplans is not enough. It is important to understand how using data in new projects will provide the necessary ROI. (return on investments)The author says that he is looking for 10x ROI for data projects.

It is important for AI and data projects to understand that such projects are complex and require skills, people and organizational changes to succeed. And this is quite difficult and not everyone can benefit from such projects. Specifically, the author says that if you run data and AI projects inside DWH teams, then such projects are doomed to failure. ("the team where good data projects go to die). This is not because DWH technology is bad, but because it is more of a human problem. ("people problem")They are very specific about problems and very specific about their work. In general, the author says that this is not the team that should be responsible for data and AI projects of a new type.

P.S. The same thing happened at YOW. 2022 I mean him. told earlier. The speech turned out well and the author re-told it at the goto conference. 2023 year:) P.P.S. I like the author’s thoughts and I planned to read his book a year ago, but still haven’t read:)

#Management #Leadership #Data #DataScience #AI #Engineering #Software #SoftwareDevelopment #ML