Everyone loves a new toy, and us folks in the tech space are no different. The hype around AI has been running at full throttle since November 2022 when ChatGPT launched. According to Gartner, we have passed the peak of inflated expectations. The challenge for businesses at this point is to realise why companies enter the Trough of Disillusionment and how to break through to the nirvana stages of Enlightenment and finally the Plateau of Productivity.
From our results published in Searce’s 2024 State of AI report, we see that 93% of organisations surveyed have seen positive benefits on their AI projects, and 25% are planning to spend 50% more on AI, with 8% allocating 100% or more, showing us they are committed to reaching that desired end state.
How to scale for the future in 2025
With any new, inherently versatile technology, the drive to adoption is hard. It takes planning to determine exactly which of the many potential benefits should be tackled first. The first people to see a flashlight didn’t have any doubts about what it was useful for, they simply “saw the light” and adopted it. Since AI offers a wide, diverse range of use cases that can both augment and in some cases fully replace tasks that humans do, it requires identification and then prioritisation of the use cases that will generate the highest ROI.
To calculate the ROI, a clear view of the current costs is important. Internal research and content creation are common use cases that AI can solve. Businesses must first understand how much internal research is being done per user group population and then derive the cost. This is where companies hit the first hurdle, as many will not have quantitative or qualitative data on how internal research is done and for how long.
Let’s take a simple example, taking a sales team of 10 people, each meeting two new customers, twice a week. For top performing salespeople, preparation is key. This includes researching the customer’s company, reading annual reports, understanding their market challenges, analysing competitor activity, considering the role of the person they’re meeting, and identifying the key metrics that matter to them. After gathering all this information, they’ll create a meeting plan with insightful questions to engage with the customer and uncover potential leads.
Let’s estimate this preparation takes about 1 hour each, that’s four hours per week for each salesperson. Now, if we augment this process with AI, those four hours of manual research are saved. Over the course of 47 working weeks (factoring in five weeks of holiday), that’s a total of 188 hours saved in a year, or roughly a 10% efficiency gain.
That time saved can be reinvested in more customer meetings. Where the sales team would typically meet 940 customers annually, they can now meet 1,034, thanks to the additional 10% capacity. The ROI for AI is evident – not only through increased sales but also in the tangible benefits. Naturally, we would see an increase in sales, but also the salesperson’s role becomes easier, with AI automating the tedious task of reading through lengthy reports and distilling key insights, allowing them to focus on what they do best: building relationships and closing deals.
Another clear benefit is that augmenting research this way ensures less experienced or less prepared sales professionals are better equipped, leading to improved performance and increased sales bookings.
While this use case was easy to understand and straightforward to calculate, organisations are required to discover and document each of their use cases within the organisation and then prioritise. Many businesses lack this level of detail on internal processes and procedures and will have to understand the current state, or “as is”, before planning for the future state powered by AI, or the “to be”.
Harnessing internal data for AI
Unlike the previous use case relying on external public data sources, content creation requires CIOs to harness internal data for effective AI integration. The amount of training data needed for an AI model depends on the application type factors like domain specificity, data quality, and diversity. For example, a chatbot will have a high degree of domain specificity, high data quality, medium data diversity and thus only a small to medium sized data set, whereas a creative writing assistant requires a much larger data set.
Back to our sales team. Let’s say we want to use AI to create sales proposals, and a good proposal will have an executive summary, client needs analysis, proposed solution, benefits and ROI, an implementation plan and finally a pricing section with terms. The CIO must consider the following:
- What types of data are most relevant? This might include past sales proposals, customer information, product descriptions, market data, and competitor analysis.
- What level of granularity is required? Should the AI model have access to all sales proposals or is a more aggregated view sufficient of just the best examples across whatever sales domains they have?
Careful consideration should be given to how data is collected and processed to comply with relevant data privacy regulations (e.g., GDPR, CCPA). As well as anonymising or redacting personally identifiable information (PII) before training the AI model.
Data can then be centralised and extracted while ensuring the data is clean, consistent, and in a format that the AI model can process.
While this gets the company to an AI model that will save considerable time in proposal creation, with more consistency and higher quality, upskilling of the users and managing this change to how people work are also key pillars of the ultimate success and ROI. People may naturally be resistant to this change and even see it as a threat to their roles.
Overcoming human resistance to AI integration
As illustrated by the use cases above, the technological aspect of AI implementation is often less challenging than the organisational integration with the people who will use it. AI’s versatility makes it a powerful tool for solving various business problems.
However, to avoid the “Trough of Disillusionment”, organisations must prioritise the seamless integration of AI with existing human workflows and processes. This requires a deep understanding of current practices—what, how, and why—and a commitment to providing comprehensive training and support. By addressing the human element of AI adoption, businesses can pave the way for wider acceptance, achieve anticipated ROI, and ultimately reach the “Plateau of Productivity” where AI’s full potential is realised.
Remember, successful AI integration is not just about the technology; it’s about empowering people to work effectively alongside it.
Julian Mulhare
Julian Mulhare is Managing Director EMEA at Searce. A senior technology executive with expertise in driving innovation, Julian has substantial experience in scaling teams by creating and executing strategies that increase revenue, enhance customer satisfaction, and foster engaged teams.
