I attended a virtual conference where a pretty interesting statistic was shared: 75% of AI & ML projects fail or benefits will not be realized. Gartner, in 2017 and 2019, seem to echo this sentiment: 80% of analytics insights will not deliver business outcomes through 2022 and 80% of AI projects will “remain alchemy, run by wizards” through 2020. Given the hype and number of businesses investing in data, the risk for negative ROI is alarmingly high. For small consultancies like Conaxon, this is not good news given that our goal is to create opportunities that allow small-to-midsize businesses to cash in on the benefits of AI and ML.

Here are the top 5 ways to drastically improve the success of AI and ML initiatives:

• Talk to Stakeholders and Include them in Decision Making:

  • A general lack of understanding surrounding AI, ML, Business Intelligence, and d Data Science can make well intentioned projects dead on arrival. Naturally, we are uncertain about new technologies, change, and being left behind. These are all valid! But, if business and data science leaders spend time educating, socializing, and strategizing how data literacy gets weaved into the company culture. If your employees are in constant fear that AI and ML are going to be replacing them then it will be incredibly difficult to allow for integration. AI and ML are not going to automate away everything. AI and ML is a tool to be used in symbiosis. These are tools to make human functions more precise and efficient.

• Start with Decision Intelligence:

  • Do not get caught up in the shiny gem that is data. It is so easy to overdo it early in the game. Start simple with AI and ML. Applied AI and ML are not yet advanced enough to easily interpret chaos. You need to collaborate with the various business functions and decide which decisions could be better by having a piece (or pieces) of information—the more repeatable, the better. AI and ML work best when the thing you are trying to make more efficient is repeatable and a pattern can be taught/identified. If the project does not meet those two very basic criteria, then your risk for failure increases fairly exponentially.

• Keep it Simple:

  • Don’t try and boil the ocean. Data can be overwhelming as well as liberating. Stay focused on a few initiatives that truly help make your team’s life easier. Putting dozens of dashboards with multitudes of charts and KPIs in front of executives isn’t effective.

• Spend a majority of the time on defining/measuring the problem:

  • If you start off your journey with ML and AI with a poorly defined and un-measurable then failure is imminent. Aimless, or poorly aimed, AI and ML development will result in the output being vastly different than what your stakeholders need. At the end of the project, your shiny new data product needs to be a tool that people use and integrate with—like a sword and an arm. Swords were an extension of the warriors arm. AI and ML products need to be integrated in the same way. As mentioned above, engage with the end-users early in the project. Interface with them regularly to assess . Study how they work day-to-day.

• Put a good team together—with a kick-ass project manager:

  • The team you build around your data vision will be the keystone for success. Your decision maker should be an advocate and ambassador. They should be pragmatic. They should have solid domain experience in the space the team is operating within. You should probably find a customer champion. This person should be politically savvy and have very intimate knowledge of how the operations are performed; furthermore, is well respected by the end users. Of course, you need your data scientists, engineers, and analysts. Last but not least, spend some time and money on finding a really great project manager. A great many analytics, AI, and ML do not come to fruition because of project management related issue. This is not to say the project managers are all to blame! However, there is something to be said about the impact of a great project management professional on the outcome of an initiative.