AI

Top 5 Problems With the Business of AI: Vendor Call To Action

In one corner: AI market forecasts: 30 – 45% annual growth rate over the next five years? In the other corner: Hiring data says most firms are not building AI development teams. Surveys of enterprise executives find that very few are significantly investing in AI (great overview here).

Businesses can certainly buy AI as part of application product vendor offerings. However, this is not really “buying AI” but buying apps that may have an AI component. Over time this will not differentiate them. Customers are not buying AI tools and products.

Whaaat? Why not?

There is a problem with the business of AI.

1. Execs Do Not Know What AI Is Because AI Is Everything

Everybody wants to be AI. One recent AI Adoption survey includes everything – RPA, chatbots, machine learning, and autonomous driving. Many technology firms cite AI as part of their technical underpinning. There is even currently an analyst firm shootout to coin even a new buzzword that integrates AI with many other technologies – hyperautomation (is there any term can be used after this on expires?), digital process automation, and intelligent process automation.

Business execs I talk to are confused – what is AI?

2. Vendors Do Not Enunciate the Value of AI in Business Terms

Why are there so many cloud business financial value calculators and none for AI? Many vendors sell the process, not the results. Or qualify results “it depends on your data”. When business results are mentioned, it is often difficult to extrapolate the results to the client organization.

One difficulty – when are results good enough? If a text analytics or machine learning engine “works” 50% of the time – is that good enough? Can it be used by the business? What about 80%? Will this percent change good or bad over time as real-time data shifts away from training data? When can a user “trust” the results?

A second difficulty – AI results are dependent on specific technology capabilities, an organization’s data, and the skills of the development staff. This means that in addition to the core AI product costs there are often significant data normalization and integration, personnel, and development costs.  These costs can be hard to explain to business executives. Most importantly, a significant financial return is required to overcome all these startup costs.

I could go on but you should see the point.

3. Sales and Marketing Focuses on Technology or Business Vision, Not Business Value

AI is often positioned as a technology or a vision, not a business solution. Where is the value proposition? A quick survey of vendor websites shows that they are mostly selling to technologists. That may be OK for the web but when the sales force does the same thing it is a problem.

Sales and marketing need to better target business executive perspectives by quantifying the business and financial value of AI to an organization. AI marketing collateral and sales decks should identify business revenue increases, cost savings, and/or other metrics that business executives use to make decisions. Show how a business will be better.

Vendors should differentiate their products and solutions from the competition by use cases and quantified business value.

4. Perception By Executive They Are Buying a Research Project Not a Business Solution

It is often difficult for business leaders to understand what they are buying. Is it a tool? Tools and data? Tools and a process? Consulting?  A “black box”?

Business execs want to pay for results. Often it seems to business executives that they are paying for a research project – with a planned but not guaranteed value – rather than a specific solution. Consequently, many execs will want to wait until someone else tests out the concept and proves the path to business value.

As well, businesses want to have low or minimal risks in their business. However, some AI technologies such as machine learning can have high or unforeseen risks. The combination of a lack of direct business benefits with high perceived risk causes many executives to slow-walk AI projects.

5. Buying AI Is Hard

By this I mean that a business can usually “try-and-buy” newer or unproven technology, buy a starter kit, or some other way try out the technology before spending a lot of time, effort, and money. It may be that AI inherently takes a lot of time and effort (equals money) to test out, regardless of whether a vendor offers a sweet deal to start.

But buying AI needs to be made easy. This is changing, but needs to be better. AI needs to be made easier to build and therefore easier to buy.

What Vendors Can Do

Vendors can do a better job of thinking like a business executive:

  1.   Lead with business cases showing financial value.
  2.   Create try-and-buy solutions or solutions-in-a-box. Make AI simpler to make buying simpler. Something that  quickly shows financial benefits.
  3.   Build and demonstrate case studies.
  4.   Invest in training. Make AI specialists more common, and consequently require less investment than is now required. Or build products that do not require AI specialists.
  5.   Be honest. Explain how you will manage the downside risks.

AI vendors need to think strategically. Two technology adoption analogies come to mind –  data warehousing and cloud computing. When first introduced both of these technologies were marketed and sold from a technology, not financial perspective. They were niche products with great potential but unclear financial benefits. It was only when financial value, ease of use, and key industry use cases were enunciated that the technologies took off.

Vendor Challenge: Sell the business value of AI!

Peter Brooks

Peter is an analyst and consultant focusing on quantifying the business value of technology.

Disclosure

The views and opinions in this analysis are my own and do not represent positions or opinions of The Analyst Syndicate. Read more on the Disclosure Policy.