A 2018 Databricks trend report conducted by CIO/IDG Research Services found that while almost 90 percent of companies surveyed are investing in AI solutions, most aren’t getting the results they need. The report states that “only one in three AI projects are successful and it takes more than six months to go from development to production.”

The fact that the overwhelming majority are investing in AI solutions is a strong indicator of the technology’s potential. But the dilemma of how to get to value quickly remains. It’s not uncommon for businesses to have issues generating value immediately from new technologies — similar delays between adoption and return on investment have occurred for many of the technological advances that have changed business operations over the past few decades.

But one factor that makes this gap unique are the reasons behind it. Unlike previous technologies, many companies focus on AI alone as an unparalleled business problem solver. A singular focus on AI isn’t completely wrong because AI is critical for business success in the modern economy. But while AI is necessary, it’s not sufficient in and of itself, and that is a crucial point that too few adopters understand.

Getting Results Faster

The good news is that it is possible to successfully deploy AI and start generating results quickly. It starts with a realization that AI alone doesn’t solve the company’s business problems. Companies need a strategy to augment AI so that the overall solution not only begins to deliver results on the original business problem it was deployed to solve but generates value throughout the organization. The strategy is called A.C.T., and it contains three elements:

Analytical Framework: Rather than simply trusting AI to come up with solutions for all their business problems, companies need to provide an analytical framework in which AI can operate. The easiest way to do that is to think of a business problem to solve and identify the data needed to solve the problem. This simple exercise is extraordinarily useful because it provides those who engage in it an understanding of how AI works to solve problems.

For example, imagine a scenario in which a company deploys AI to help their sales and marketing team predict when customers are likely to make a purchase. To solve that problem, they’d gather data on where prospects are in the customer journey, take a look at historical purchasing data, etc. AI needs that same data to deliver results, and lack of an analytical framework hinders returns on the AI investment. The Databricks report notes that “data-related challenges are hindering 96 percent of organizations from achieving AI” goals.

Context: AI can learn much from data, but it doesn’t automatically pick up on business context. That’s because most datasets don’t contain sufficient information to determine context, or relevant context simply isn’t included in the data available. That means business users need to provide context to ensure that the AI solution fulfills its potential. For example, humans know new products are generally more appealing to customers than established products, but AI won’t know that unless a user tells it. Hence, AI, without context, may bias recommendations to past popular products.

Context like customer budget information, specific price points that appeal to customers, and other business rules must be conveyed to the AI solution before it can deliver the best results. Too many AI innovators make the assumption that AI knows this information already, but it only factors in the context it receives or learns on its own, so it’s important to ensure that relevant context is conveyed to the AI solution.

Technology: The last piece of the puzzle is technology. Many companies gear up for an AI implementation by building AI, machine learning and data science teams, but they continue to use traditional technology. That’s a mistake because they need a model that is scalable. However, because technology and analytics teams tend to be separate, companies often miss out on the opportunity to scale AI learning throughout the organization.

To generate the most value from an AI solution, it’s a good idea to leverage AI-compatible tools that the company already uses, such as Salesforce. Solutions created by the AI community are also worth a look — and much more efficient than building a solution in-house. Point solutions can’t scale and are therefore usually a hindrance rather than a help. For these reasons, a scalable platform is recommended.

By using AI with the A.C.T. strategy, companies can solve the AI dilemma and generate value much more quickly, reaching the goals they had in mind when they implemented AI. But perhaps even more importantly, the AI + A.C.T. strategy can leverage AI learning throughout the enterprise, generating results across business units and delivering value far beyond those original objectives.