Reading time: 2 & 10 minutes

Welcome to the Tech A newsletter DeepDive
The reality is that despite massive investments in AI during 2023-2024, only 26% of businesses are generating tangible value. The good news: it’s not just hype and you can generate positive ROI from AI adoption across your business. In fact, those who succeed are seeing unprecedented returns - nearly 3x ROI and 30% premium on key financial metrics (Accenture)
The question? how to do that effectively with the level of AI maturity we have now (2024).

This week’s deep dive explores the real value of AI in businesses, examining the current context, key challenges, and most probable approaches to generating real impact from AI.
I’ve combined my own experience with enterprise survey results from more than 3000 companies (conducted by BCG, Deloitte, McKinsey, Dataiku, and more).

Below is Part 1 of our three-part deep dive, where we explore AI usage by different industries & business lines, most common AI techniques and return of investment numbers.
In the following weeks, we'll explore:
Part 2: Why Do Most AI Projects Fail?
Part 3: The most effective strategies to implement AI?

⚡ Your 2-Minute Summary

In this article, I wanted first to answer the questions: what do the businesses use AI for? Are there sectors or business lines that see the value of AI the most (like sales & marketing)? Finally, my most important question, are some of these businesses seeing tangible return on investment beyond the hype?

Key Findings

Current State:

• Only ~15% of companies use AI at scale; 80-85% remain in proof-of-concept stage

• Financial services, Tech/Telecom, and Industrial sectors lead adoption

• Highest ROI seen in financial services (60%+ see $1-5 return per dollar spent)

Key Implementation Areas:

• 62% of value comes from core business functions (operations 23%, sales 20%, R&D 13%)

• 38% from support functions (customer service 12%, IT 7%, procurement 7%)

• Most common: predictive analytics (85%), forecasting (79%), LLMs (68%)

Critical Insights:

• Most organizations prefer hosted LLM solutions (43% deployment) vs self-hosted (14%)

• Enterprise average spend: >$1M yearly on GenAI

• Only 26% generate significant value, with 4% being AI innovation leaders

• Successful implementations show 3x ROI and 30% premium on key financial metrics

Keys to success: Focus on core business integration rather than just support functions, with a strategy tailored to industry-specific value drivers.

How do businesses use AI?

If you’re reading this, and despite the hype, only a minority of companies are using AI at scale.

You are most probably using AI for your own productivity within your team or your IT department, which is driving the initiative. This means that you’re still in the proof of concept stage and are not reaping most of the benefits AI can give you.

A study from Accenture showed that around 80-85% of companies are stuck in this initial phase. Phase 2 is AI scaling across the organization (10-15%). Organizations in phase 3 are the few pioneers (<5%) who achieve a culture of AI with data and analytics democratized among the entire organization.

We’re in the phase where most companies are exploring and adopting AI, with only 15% of companies globally not currently exploring the use of AI.

Financial services, Technology & telecommunications, and industries are leading the way in adopting AI.

Each industry uses AI in a few different use cases, with common applications emerging like IT automation, HR and talent, and Marketing and sales. I suggest you check the full table in IBM’s report to get a list of +20 use cases by industry.

Operations, IT, Customer Service, Marketing & Sales among the departments seeing most of the AI adoption & value:

What’s interesting about the chart below is that support functions are picking up pace in the AI adoption race. From what I’ve seen, horizontal AI solutions in sales and marketing and customer experience are the most implemented solutions. That’s mainly due to their effectiveness, low risk nature, and immediate ROI benefits (time saved to generate creative content, increased customer satisfaction rate, increased sales, etc.).

You can integrate AI into these business lines for your initial pilots. Moving to scale will, however, require an enterprise-wide strategic approach with a more extended timeframe for effective implementation.

AI Technologies & Techniques: Current State of Implementation

If we delve a little bit into the AI technologies being used, we see that LLMs (OpenAI, Anthropic) are not the top-used AI techniques. In fact, predictive analytics and forecasting have been used by data scientists in various sectors for decades. What’s fascinating now is that any of the 6 AI techniques below can now be implemented in a fraction of the time it took before. The AI creation pipeline has been vastly improved with massive investment and attention.

The comparison between 2023 and 2024 reveals key shifts in AI technology adoption:

Established Technologies Show Stability:

• Predictive analytics remains dominant but shows slight decline (90% → 85%)

• Forecasting maintains a strong position (83% → 79%)

• Fraud detection holds steady at 63%

Emerging Technologies Gain Ground:

• LLMs achieve a significant 68% adoption in the first year

• NLP and image recognition show steady growth

• Image recognition sees a notable increase (60% → 66%)

The slight decline in predictive analytics and forecasting might indicate a resource shift toward newer technologies like LLMs rather than the decreased importance of these foundational techniques.

LLM Implementation Landscape

The stark difference in production deployment rates (43% vs 14%) suggests organizations prefer the lower barrier to hosting solutions entry, despite potential customization benefits of self-hosted options.

Are we seeing results of positive AI adoption?

Businesses have made considerable investments in 2023 and 2024. On average, enterprises spend more than $1M yearly on GenAI(Dataiku). The sentiment is positive across the board, and everyone wants to leverage AI to generate exponential business impact.

However, according to the numbers, only 26% of them generate value from AI, with only 4% at the forefront of AI innovation (BCG).

For those who succeed in scaling, things look pretty good, with a nearly 3x return on investment and a 30% premium on key financial metrics.

For those who do not gain significant return on investment, most of them are stuck in the proof of concept phase. According to Deloitte, 70% of respondents said their organization has moved 30% or fewer of their Generative AI experiments into production.

Financial services, banking & insurance have the highest probability of positive ROI (Return On Investment):

According to a recent report by Dataiku, most companies (+60%) adopting AI see a positive ROI between $1 and $5 gained on each dollar spent. However, as many companies struggle to measure and implement AI adoption metrics, these results should be taken with a grain of salt.

The results show that financial services, banking, and insurance see the most promising return on investment with the lowest downside. As we’ll see later, these industries also focus their AI strategy on core business processes rather than support functions.

Approximately what return do you deliver for each $1 spent on data, analytics, and AI initiatives? Adapted from AI: Today report by Dataiku

Across sectors, most of the value is in integrating AI into core business functions, not just support functions:

Based on the common feedback I receive, this was surprising for me! I thought AI would be best suited for support functions. However, this report from BCG outlines a different story.

A clear pattern emerges from successful implementations:

From what I have seen, pilot AI implementations focus on support functions to avoid potential negative impacts on core businesses. However, the leaders implement a company-wide implementation of AI into core businesses to leverage its capabilities.

The positive impact of AI on different functions/departments has a high dependency by sector. For example, insurance & fintech benefit the most from AI in customer service (24% & 18% of created value). Biopharma & Automotive industries benefit most from R&D (27% & 29% of value created). BCG.

What this means: an initial analysis and an overall implementation strategy will improve your ROI on AI drastically.

The differences by sector go beyond the core and support business value. Regarding their approach to AI, Technology, Media, and telecom emerge as the best sectors implementing AI effectively.

To illustrate sector differences, let’s look at insurance & biopharma:

Insurers focus on operations (policy administration, underwriting, and claims management), customer service, and marketing and sales (BCG).

For Biopharma companies:

💡 Looking Ahead

The next article in this series will explore why AI projects fail to scale beyond siloed experiments and what distinguishes successful implementations from the rest. We'll dive deep into the critical success factors and provide practical guidance for avoiding common pitfalls.

📧 Was this email forwarded to you? Subscribe below

If you liked this first part, consider subscribing to and sharing the newsletter. It will help me do what I like: research and write about topics related to tech and business.

Sources:

• Dataiku Survey Report: AI, Today (2024) with 400 senior AI professionals

• IBM: Global AI Adoption Index Report Dec 2023

• Accenture: AI: Built to Scale Report

• BCG: Where’s the Value in AI? Oct 2024

Amine Rabehi

Find me on LinkedIn or Book a Free Call

🔍 Whenever you’re ready, here are 3 things I can help you with:

• Technology Solution/Product consultancy: I help you assess your product idea from a technological & business point of view. So far, I have built a network of professionals covering the MENA region in term of IP, Market Research & Complex AI solution development.

• Market entry strategy and support: understand the market, legal framework

• Tech Product Development Solutions: I’ve managed more than 15 tech product development projects (hardware & software). I help you go from idea to initial batch launch.