AI for ROI: Methods to Guarantee Optimal Returns from Your AI Investment

Many IT leaders find themselves in a tough spot when trying to get approval for AI projects from their CFOs. The big question they face is: How do you calculate the return on investment (ROI) for AI? It’s tricky because while AI has immense potential, its benefits often show up indirectly and over a longer period. This makes it hard to present a compelling ROI case to CFOs who have to prioritize immediate financial returns.

However, there’s some good news. An IBM study shows that when companies carefully plan and decide where and how to implement AI, the payoff can be substantial. On average, companies that deploy AI strategically see a 13% return on their AI projects, which is more than double the typical ROI of 5.9%. But wait, there’s more. Another study found that for every dollar invested in AI, companies can expect an average return of $3.50.

At Flyaps, we’ve seen firsthand how transformative AI can be. As a software development company with a track record of successful AI implementation projects, we’ve helped organizations across various sectors, from human resources to urban planning, to harness the power of AI and help it meet their business objectives. In this article, we're going to break down how you can measure and achieve ROI with AI. We'll explore how to turn AI from a vague promise into something that delivers real, tangible results. Keep reading to find out six steps that will let you double your ROI from AI investments! 

Steps to achieve impressive ROI from investing in AI

Now, a good ROI is the result of a well-thought-out strategy. In the case of AI, you have to follow 6 steps, which we will now discuss.

Step 1. Identify where AI can be put to the best possible use

AI is considered one of the technologies that helps businesses streamline almost any business operation. But it’s not always the best tool for every job. For instance, you want to automate invoicing processes and the customer support area of your business. Though AI can be good at both, robotic process automation (RPA) can be more cost-effective and faster to adapt for invoicing automation. RPA excels at handling repetitive, rule-based tasks like entering invoice data into your system. It’s straightforward, doesn’t require extensive training on data, and can be set up relatively quickly.

On the other hand, if you want to improve how you handle common questions from your clients, AI is definitely your go-to. Specifically, using a large language model (LLM) – a type of generative AI – is ideal for building a chatbot. LLMs are designed to understand and generate human-like responses, making them perfect for conversing with customers.

AI strategy should be driven by a genuine need for AI, not just because it’s trending right now. According to research by IBM, organizations that truly understand the importance of AI to their business strategy are 1.8 times more likely to achieve the double ROI we talked about earlier. 

So overall, if you know exactly how AI can make a difference in your operations, you're much more likely to see significant returns on your investment.

Step 2. Choose an AI operating model

Once you know where exactly to put AI in place, it’s time to think of the type of AI operating model – the framework that guides how AI is developed, implemented, and managed. Just like different teams might have different strategies in sports, businesses need to decide on the best approach for integrating AI based on their specific needs and resources.

There are three main types of AI operating models:

Centralized, when AI initiatives are managed by a central team. For example, a multinational bank has a central AI team that oversees all AI projects, from fraud detection to customer service chatbots. This team sets the standards, develops the AI tools, and ensures that every project aligns with the bank’s strategic goals and compliance requirements.

Hub-and-spoke, when a central AI team (hub) collaborates with decentralized teams within different departments (spokes). Suppose, the bank's central AI team develops core AI tools and frameworks. Each department, such as retail, investment, and compliance, has its own AI specialists (spokes) who adapt these tools to their unique needs. For instance, the retail team uses AI to enhance personalized customer service and detect potential loan defaulters, while the compliance team uses AI to monitor transactions for suspicious activity. This way, the central team ensures strategic alignment and provides support, while the departmental teams tailor the AI solutions to their specific requirements.

Decentralized, when AI projects are managed independently within each business unit or department. In this case, each department within the bank operates its own AI initiatives without central coordination. In our imaginary example, the retail department would develop its own AI-driven tools for customer behavior analysis and personalized banking recommendations while the corporate banking division would be busy creating separate AI models to predict market trends and investment opportunities. While this operating model lets each department tackle their own challenges, it can also lead to wasting resources and duplicated efforts. For instance, different teams might work on similar projects, like developing their own fraud detection algorithms.

Step 3. Consider following the artificial intelligence operations (AIOps) approach

AIOps is a practice to enhance and streamline the management of IT systems using AI. Basically, if you combine Agile DevOps,  MLOps, and automated ITOps with AI you will have AIOps. The goal of this approach is to deploy AI solutions that are flexible, user-friendly, and scalable. To do this, AIOps incorporates three crucial elements:

1. Skilled teams who develop and manage AI solutions.
2. Structured processes that ensure AI projects are developed and deployed efficiently.
3. The technology and tools that support AI applications.

Together, these elements help organizations apply AI quickly and on a large scale.

When it comes to achieving better ROI, AIOps is aligning AI operating models to ensure that IT operations are managed consistently across different departments. For example, in a hub-and-spoke model, the central AI team could set up the core AIOps infrastructure and tools and the departmental teams would then customize these tools to meet their specific needs.

Step 4. Automate and protect your data

Data is the lifeblood of AI. No other technology depends as deeply on data for its success. The way a company manages its data can make or break the success of its AI projects and directly impact the ROI.

Here’s how you can make sure your data helps you get the most out of your AI initiatives:

1. Set up systems where your team can quickly access the data they need. For example, use a centralized data platform where all your important information is stored in one place.

2. Automate your data collection and cleaning processes to save time and ensure consistency. For instance, opt for tools like Trifacta Wrangler or Astera Centerprise to automatically gather needed data from different sources and clean it up, so it’s ready for AI to analyze right away.

3. Implement systems that continuously monitor and fix data errors in real-time, ensuring your AI models are always working with the best possible information.

4. Use encryption and strict access controls to make sure that only the right people can see or use sensitive data.

You can read more about data management for AI in our article “10 Principles to Create a Strong Data Foundation for AI Success.”

Step 5. Invest in a skilled workforce

According to Upwork, job postings for generative AI skyrocketed by over 1000% in just the second quarter of 2023. It shows that businesses understand the growing role of the right AI talent and skills in their organization and its impact on ROI. 

You can start by clearly defining the roles you need for developing and maintaining your AI project. It can be specialists who develop AI models, data engineers who ensure high-quality data, and others. 

A full list of the AI roles you can read in our dedicated article “9 Generative AI Roles You Need for Your Business.”

Continuous learning helps employees stay up-to-date with the latest AI advancements, enabling them to apply new techniques and tools effectively. Create all needed opportunities for that and, additionally, provide employees with no-code or low-code modeling tools that simplify AI development.

Step 6. Be open for change and co-creation

The real value of any AI system unfolds only when employees and stakeholders trust such tools. AI adoption isn't always smooth sailing. Think back to when the Internet just became mainstream. Many were skeptical, but today, it's an integral part of our lives. Similarly, for AI to be embraced fully, employees and stakeholders need to believe that it's worth adopting.

To build this trust and drive adoption, you can take several steps:

• Launch small, manageable AI projects to test the waters. Such projects help teams understand AI’s potential and build confidence before rolling out larger initiatives. For example, a small marketing firm can start its AI adoption with an AI-powered email campaign tool that analyzes historical data and recommends optimized subject lines and personalized recommendations for improving the open rate. If the project proves to improve the marketing team’s efforts, the team will be convinced of AI's capabilities. 
• Use AI tools like ChatGPT during meetings or brainstorming sessions to showcase its practical benefits and encourage others to get on board. 
• Identify and support enthusiastic employees who can advocate for AI. Suppose, a tech company provided a software engineer with a keen interest in AI with additional training and resources to deepen their AI knowledge. Over time, the engineer may start holding workshops for their colleagues, sharing insights and successes with AI projects. 

Key KPIs to calculate ROI

KPIs can be direct, such as cost savings from automation or revenue increases from AI-driven marketing campaigns. They can also be indirect, capturing qualitative benefits like improved decision-making or competitive advantage.

Examples of direct KPIs:

Cost savings. Measures reductions in labor costs, improved operational efficiency, or optimized resource utilization achieved through AI automation.

Revenue increase. Lets you track gains in revenue from enhanced customer acquisition, cross-selling, or personalized recommendations driven by AI insights.

Marketing effectiveness. Helps evaluate metrics such as cost per acquisition (CPA), conversion rates, or return on advertising spend (ROAS) from AI-powered marketing campaigns.

Examples of indirect KPIs:

Improved decision-making. Suppose, a retail company uses AI to analyze vast amounts of sales and customer data, leading to more informed and strategic decision-making about inventory management and marketing tactics. This results in better alignment with market demands and more effective promotional campaigns, even if the direct financial impact isn't immediately clear.

Competitive advantage. For example, a logistics company adopts AI to optimize route planning and delivery schedules. While direct benefits like reduced fuel costs are apparent, the long-term competitive advantage gained through faster delivery times and improved customer service positions the company as a market leader, attracting more clients and partnerships.

Though indirect business factors will definitely influence your AI’s ROI, they are challenging to calculate. With that said, let’s look at how to calculate direct KPIs.

ROI calculation formula

Basically, you can calculate the ROI of your AI project in three steps:

1. Start by determining how much money the AI project has saved by reducing or eliminating certain expenses. This could be from automation, improved efficiency, or reduced errors.
2. Next, measure how much additional income the AI project has generated. This might come from new sales, enhanced customer acquisition, or other revenue-driving activities enabled by AI.
3. Finally, add up all the expenses related to the AI project over its entire lifecycle. This includes initial setup costs, ongoing operational expenses, maintenance, and any other costs associated with running the AI solution.

Once you have these values, plug them into the formula: ROI = cost savings + revenue – total cost of ownership (TCO).

Let’s say a company implements an AI system that:

1. Saves $200,000 annually by automating repetitive tasks (cost savings).

2. Generates an additional $300,000 in revenue through improved customer insights (revenue).

3. Has a total cost of ownership of $250,000 over its operational period (TCO).

Plugging these values into the formula: $200,000 + $300,000 - $250,000 = $250,000

In this case, the ROI is $250,000, meaning the AI project has generated $250,000 after accounting for the costs of implementation and operation.

Real-world example: how GitHub Copilot enhances ROI 

$21 per user per month can be a significant investment if you have a big team of developers. That is exactly how much GitHub Copilot’s enterprise plan costs. 

GitHub Copilot is a generative AI tool designed by GitHub to assist developers in writing code. But according to Kyle Daigle, GitHub's Chief Operating Officer, the vision behind Copilot goes beyond simply assisting with code generation. The goal is to help developers build impactful projects more efficiently, driving broader human progress. Copilot reduces the time to market for new features and products by accelerating development cycles. This time efficiency translates into cost savings and revenue growth. But let’s do some math. 

Suppose, the average cost of a software developer is $120 per hour and you have a team of 10 developers using GitHub Copilot. Let's assume a developer works 160 hours per month (40 hours per week). The monthly cost for 1 developer will be $19,200 (160×$120). Monthly cost for 10 developers - $192,000 (10×$19,200). 

According to GitHub's research, their Copilot helps developers write code 55% faster. When developers are 55% faster, they complete tasks in 45% of the time. With the productivity boost, each developer effectively works 72 hours to produce the same amount of work (160×0.45). The monthly cost per developer using Copilot will be $8,640 (72×$120), and the monthly cost for 10 developers will be $210 (10×$21).

The resulting monthly savings will be $105,600 ($192,000-$86,400). The monthly cost of Copilot for 10 developers - $210 (10×$21). Finally, the net monthly savings is $105,390 ($105,600-$210).

While it's tough to put a number on the improved job satisfaction and well-being that GitHub Copilot as an AI tool brings, the concrete financial benefits are clear as day. Therefore, by looking at the direct savings and increased efficiency, companies can see immediate returns on their investment in AI. Not only that, they can also forecast future financial gains from adopting AI tools like this.

Ensure higher ROI with Flyaps today

When we at Flyaps create AI solutions, we always put your business goals and financial ambitions at the center of the development strategy. We don't just aim to fix your immediate problems; we also design tools that can become valuable assets. Like it was with CV Compiler, for example, when we built the AI tool that not only streamlined our client’s processes but also became a popular tool for other businesses facing similar challenges. 

Moreover, we have a range of pre-built AI tools ready to go. You can easily integrate them with your systems and enjoy the benefits of AI without the high investment usually associated with building everything from scratch. 

Looking for an experienced team that will help you not just solve today's issues but also gear up with powerful tools for the future? Drop us a line!