We are living in the computer age and entering the era of artificial intelligence. No major company today can live without software. With the advent of AI technologies, the need to secure a company’s operations with a quality system is more urgent than ever. AI can already accelerate the work of humans, in some cases replacing human work altogether. The benefits of implementing AI-integrated software are very appealing. For those already working on it, it will provide a competitive edge that can be a decisive factor for a company’s market position within a few years. As this can be a significant investment, it is worth taking some time to calculate the return.

What is the dependence of your business model on software?

The extent to which the investment is significant can be obtained by answering the question of how much the system affects the performance of the company and how much the company can live without it, or how much the software is substitutable by alternatives. There are fewer and fewer industries and professions untouched by AI. With the advent of radical innovation, lists of professions that new technology will replace used to circulate. Now there is an ever shrinking list of professions that will not be replaced. These include, for example, motorcycle repairman. So, if you repair motorcycles, for example, you are fine. If not, you should begin to wonder.

How deep do you need to go in AI?

Artificial intelligence can be just a parametric improvement to a company’s existing processes or it can become the vehicle for an entirely new or innovative business model.

The significance of an investment in a system can be rated on a scale of 0 to 1 in terms of:

0 … we only use easily replaceable boxed office software
0.5 … we use process management software, the system is easily replaceable by alternatives
1 … we manage the company’s core processes with customized software or a self-developed application

How dependent the operation of your firm is on the investment being evaluated will answer which method to use to evaluate the return on investment.

For implementations with a low impact of the investment on the functioning of the company (on a scale of 0 to 0.5), a simple ROI calculation, payback period or payback analysis can be used. The goal is to minimize payback time and risk.

For investments with higher significance (on a scale of 0.5 to 1) it is appropriate to evaluate the benefits of the project using NPV, IRR, EVA, alternatively ROI, TCO or CLV can also be used. A glossary of terms can be found at the end of the article.

Sky is the limit

Gone are the days of linear market development. Sam Altman, CEO of OpenAI, the company behind the most successful AI tool to date (ChatGPT) says that AI tools will enable independent entrepreneurs to create a $1 billion business. In an extreme case, one can imagine a set of automated machines created by a skilled programmer/businessman and visionary that will generate revenue on their own.

The scale at which AI tools can be used is infinite and, for example, a formula for calculating the present value of the return on increasing perpetuity (perpetually increasing revenue) will not provide meaningful information, since the present value of such an investment is an infinite return.

How to reduce the risk from a wrong estimate?

Create a set of scenarios and simulate the impact on the outcome. That is, what happens if… for example, the input parameters change. Software development or deployment becomes more expensive, deliverables don’t arrive as quickly (typically software projects take longer than the entrepreneur would like), outputs aren’t delivered instantly in the ideal volume, etc.

Based on the scenarios, the entrepreneur has to draw a conclusion about the riskiness of the investment and decide on the investment based on his/her attitude to risk and the availability of funds.

Which AI projects will be successful?

Essential for the successful implementation of AI software is permanent validation and improvement, in a word “pivoting”. Those who can be flexible will win. They pivot faster than the competition and can learn and implement changes quickly. Methodologically, this process can be compared to the PDCA method. You set a hypothesis (P), make a prototype to deploy to the market as soon as possible (D), verify how it works / how customers react to the offering (C) and draw corrective actions from the verification (A). Flexibility is therefore of great value in AI deployment projects, provided the company knows how to use it.

The time value of money vs the price of indecision

Classical financial models account for the time value of money. Today you invest your own or borrowed money, the owner of this finance expects a profit, some of the income is consumed by inflation. But with the advent of AI and the acceleration of innovation, the more significant cost is the price of indecision (at one extreme) and the price of rashness (at the other). At these extremes, we need to know how to balance. Invest in the proper flexible technologies with the prospect that our team can implement them effectively and build the ability to make quick decisions about changes in the management of the business. The later you make the decision, the later the results will come. In addition, faster competitors may eliminate your efforts to generate profits from AI implementation altogether.

What are the challenges of measuring the return on investment in AI?

Unmeasurable benefits

Some of the benefits of introducing AI into a company’s operations are unmeasurable. For example, a company’s ability to make better decisions, increased customer satisfaction or retention. These are key parameters, but cannot be easily quantified.

Initial cost vs learning curve

The initial investment will be influenced by the level of existing digitalisation of the company. If a company does not have a good foundation in information technology, the costs can be enormous. If the firm is well equipped already and existing applications (such as ChatGPT) are used for AI integration, the upfront cost may not be significant. The initial rollout of an AI project itself tends to be quick and accounts for 20% of the cost and effort (time and money). The key cost (80%) arrives over time when the model needs to be refined and tuned. Benefits, on the other hand, will come later in the project (when the company has mastered it). However, then the firm may become unstoppable.

Rapid changes in technology

The challenge of calculating the return on investment in AI systems is the unpredictability of technology adoption. The tools you may have been developing and integrating a year ago will be obsolete by now; new tools will arrive that will need to be adapted quickly and appropriately.

Calculating the impact of AI on business operations

Calculating the change in labor productivity or increase in sales volumes due to AI may not be easy, as people’s job descriptions may change overall as the project is deployed. The advantage will be realized by companies with creative people at the core with the ability to make the most of AI tools in their work. A small team of skilled people will be able to start competing with the giants. A large number of people in a company will no longer be a measure of a company’s competitiveness.

Unpredictable outcomes

The nature of AI projects is largely experimental. Only with a team of skilled people who will be able to react quickly to changing environments and conditions (both in technology and market) can a project be successful.

Where to go from here?

The low ability to predict the outcomes of AI projects should not be a barrier to trying. Evaluating AI-driven projects should not just be a matter of finances, but also a corporate strategy that values the benefits – e.g. maintaining market position, opportunities by creating a new product or service, retaining and self-actualizing quality people in the company, etc.

In order to evaluate these projects, it is useful to create scenarios that may occur during the implementation of the projects. Will it be a total fiasco? How many times will it be necessary to reinvest the amount of money intended for implementation? Are permanent investments in research and experimentation a stable part of the corporate budget?

Decide on your level of innovations

If you decide to go down the innovation route, I recommend setting aside a portion of your company’s turnover to be invested in strategic development, similar to how you determine what % of your turnover you dedicate to marketing. In the future, organizations will be measured more by the level of innovation and investment devoted to automation in the form of AI and the budget invested in marketing, rather than the number of people.

Cost Benefit Analysis

Download template here, adjust it’s contents for your needs and ignite your AI journey …