Are you finding it harder to get management buy-in for AI projects? Here are 4 tips that will help you make a great business case.
More businesses today are exploring AI technologies and tools for process automation and improving productivity within their organization. A Gartner survey has shown an increase in the percentage of organizations that have deployed AI to 19% in 2020 and 24% in 2021.
The most successful implementations result in clear benefits, whether it be a visual application such as a Computer Vision-powered solution that can recognize faces, or an omni-channel one like an NLP-enabled chatbot that can automate conversations with customers.
But AI adoption is much more than just these technologies and modelling techniques. Often, organizations jump into large AI implementation exercises only to stop at a pilot stage, call it a failure and abandon the use case prematurely before realizing the huge potential down the road. It is therefore natural for business leaders to have some level of healthy skepticism when it comes to investments in AI.
Why is it challenging to make a strong business case for AI projects and what can be done to address them? Let us look at three key factors that complicate things.
AI projects generally take time to realize ROI. So it can be hard to make a business case when organizations cannot predict the costs or benefits.
For some AI applications, it may be easier to calculate ROI. For example, an energy producer that adopts an AI-powered predictive maintenance tool could track equipment uptime and maintenance costs as a way of measuring returns on their investment.
Other applications can be more complex and unpredictable. For example, reduction in patient waiting times at a hospital could be due to a number of factors such as better triaging, information dissemination, increased patient satisfaction, faster customer support and right-siting.
One of the reasons that organizations hesitate to go full steam with AI solutions is the lack of talent availability. One recent Gartner survey showed that about 20% of organizations find skills availability to be one of the biggest obstacles to deploying AI.
If business leaders are not confident that they will have the necessary talent to manage AI projects all the way until it generates significant value for businesses, they are not likely to make serious investments in the program.
Another big factor is the data volume and complexity. The algorithms and models used in AI projects make a big difference for sure but almost equally important if not more is the interactions between the data and algorithms. The amount of effort required to collect, analyze and perfect data for AI is not insignificant, and how you approach this will have a lasting effect by being usable for building many models and for multiple organization-wide use cases.
The amount of effort required to collect, analyze and perfect data for AI is not insignificant, and how you approach this will have a lasting effect by being usable for building many models and for multiple organization-wide use cases.
In fact, even if your team lacks experience with algorithms, you can still make use of packaged applications or APIs but if you do not have the data and domain expertise, you will not be ready to start any meaningful AI applications.
So how can change leaders and digital transformation drivers overcome these obstacles and articulate a clear business case and rationale to invest in AI technologies to top management and decision makers? Here are some pointers to keep in mind.
It is important to approach AI projects with a long-term view. ROI calculations should take into consideration your organization’s stage in the digital adoption journey, management culture and the seriousness in investing.
In the initial stages, the project can seem costly without any immediate gains, especially if the use case is entirely new and you are not used to setting aside a budget for these.
Change leaders should analyze the necessary investments, including that for technology, implementation, integration and maintenance. It is also important to keep in mind that these costs can change considerably as the solution scope is explored and refined. Of course, when experimental AI projects do not show signs of showing sustainable benefits in the long run, you should also be willing to close them down.
No matter how fancy the AI technology is, decision makers need to know how it moves the needle when it comes to business goals. Gartner studies show that 39% of the successfully deployed AI projects included a financial analysis on risk factors or conducted ROI analysis compared to 41% that did not deploy AI projects and only looked at the number of projects developed.
To find out what business value the solution will create, you may need to go beyond simple financial metrics. You may need to highlight the metrics of the AI project as well and how it indirectly affects the bottom line.
For example, in the case of an AI chatbot for an insurance firm, you could include the financial benefits such as cost savings from handling general queries and FAQ, productivity uplift from new business acquired by consultants who use the chatbot. You could also look at total queries managed and successfully resolved by the bot without human intervention, the reduction in call volumes and average call duration at the call centre. Tie these back to productivity and man hours saved when you prepare the business case.
Also, be sure to include a list of intangible benefits that are very important, such as:
As discussed above, having sufficient data to train the AI is a prerequisite for success. In fact, data preparation happens to be one of the foundational features essential for chatbots and conversational AI platforms.So when you are preparing the business case, ensure that the problems and use cases you are tackling have sufficient data available. The data should contain patterns that executives expect to see in the future. For example, if you are expecting a seasonal impact, your data should span multiple years to be able to represent seasonality considerations.
If you do not have the data readily available, start putting in place measures to collect and label or categorize them. A good place to start is to talk to the more customer-facing teams in your organisation who interact with customers daily.
Change leaders should prioritize training of staff on data science tools as part of the business plans. It is a good practice to have a governance body that builds up to become a shared resource centre. This team can work with relevant stakeholders including vendors at the start. Over time, they could also morph into a centre of excellence within your organization, capable of building your own solutions if needed.
It is also critical to put in place clear roles and responsibilities when it comes to managing AI projects. This is especially true of chatbots for example, where the success of the AI bot depends on constant tuning and training of the bot to improve its performance over time. In addition to your IT personnel who will be in charge of maintenance, integrations and upgrades, your non-technical team should also be trained on how to add, edit or update intents and entities and track the chatbots analytics. In the event of staff churn, it is also important to ensure continuity of such talent with clear business-as-usual (BAU) plans.
To learn more about how you can implement AI chatbots in your organization, ask us for a free demo.