How to master value realisation with your AI Customer Agents

Value realisation is the toughest challenge for succeeding with AI Customer Agents today 

Deloitte predicts that by the end of 2025, 25% of companies using GenAI will launch AI Agents pilots or proofs of concept. This will only increase by an expected doubling in use of AI Agents to 50% by 2027¹. 

Companies are eagerly seeking use cases to implement AI Agents and quick wins to drive short-term success. However, while the benefits of such solutions are a hot topic, realising actual value from AI Agents remains one of the biggest challenges for successful implementation. With only 11% of Nordic organisations deeply engaging with AI Agents, the struggle to achieve true value at scale is real².  

When looking into AI Customer Agents, the challenge is even more pressing, as the impact of “not getting it right” is even higher. Not having a clear value linkage, starting with the wrong choice of technology, and lacking an overview of high-value use cases, will leave leaders with a high risk of wasting CAPEX on use cases that never become realised in their global organisations. 

When we talk about value realisation, we refer to the end-to-end process for identifying, implementing, monitoring, and evaluating value in AI Customer Agent implementations. Value is often defined in the beginning – e.g., via a business case -  and expected in the end of a project – e.g., via a report with ROI.  

Yet, value should be treated as the natural evolution from business objectives and business case – to deeper levels of insights that continuously inform about customer desirability, technical feasibility, and business viability. 

According to Deloitte's recent State of the Generative AI in the Nordics Q4-report, Nordic organisations are exhibiting declining trust levels (-13pp.) and rising uncertainty in GenAI (+12pp.)². This underlines why value realisation plays a crucial role, as it helps drive stakeholder buy-in and prove business value, ultimately securing ROI. 

Getting started: Take a holistic approach to value realisation linking business objectives to KPIs

Value realisation is essential for developing a customer-centric solution and proving your AI Customer Agent in your business while minimising organisational resistance. But how should leaders approach it? We propose working with value realisation through Deloitte Digital’s Value Blueprint, more specifically: 

It sounds like simple logic, yet establishing this type of value linkage is often forgotten or left as a referral to a business case. The benefits of specifying the Value Blueprint with a clear link from business objectives to KPIs are: 

• Knowing the most valuable use cases of your AI Customer Agent based on customer needs, business value, and technical development. 
• Having a clear business-led view and ability to set and communicate clear business targets. 
• Defining the boundaries for what the solution will achieve from a business perspective, helping align stakeholder expectations and expected ROI. 
• Providing clear direction for development teams towards specific business and customer experience KPIs. 
• Building compelling “proof points” and change narrative for stakeholder buy-in and strategic decision-making. 

Working on the Value Blueprint could output a holistic value overview of your AI Customer Agent, such as in the illustrative example. In the following three sections you will learn the initial steps towards 1) Identifying the most valuable use cases, 2) Setting success criteria and 3) Defining KPIs for your AI Customer Agent.

1) Start by identifying the use cases of AI Customer Agents that bring value

The first step in defining value is identifying the use cases of your AI Customer Agent that enable your business objectives, e.g., boosting top-line growth or increasing internal efficiency.

The way to think about use cases depends on the business objective you aim to solve through the AI Customer Agent. For example, if the objective is to boost top-line growth, you should consider use cases related to the early funnel of the customer journey, e.g., smart search to navigate complex product inventory landscapes (endless isles), or use cases related to lead generation to increase the lead conversion rate.

If the objective is to increase efficiency, you need to think about the “jobs to be done” by the specific unit(s) that you want to optimise. For example, if the objective of the AI Customer Agent is to increase efficiency for the sales team, then you might need a chatbot to communicate with customers, a product recommendation tool for personalised suggestions, and a quote tool to design personalised customer quotes.

Use cases represent the concrete jobs of your AI Customer Agent that realise business and customer value, acting as a guiding principle for product development. In this context, attempting to achieve everything at once is often ineffective. Especially for AI Customer Agents where the value realisation can be uncertain, organisations must strategically plan and prioritise use case development  through a phased approach. 

2) Choose success criteria that tie strategy and AI Customer Agents together

When you have defined use cases, the next step revolves around setting up the success criteria for your AI Customer Agents. Let us explore what good success criteria might look like, key considerations when defining them, and what challenges might arise in the process.

Defining success criteria means identifying the core purpose of the solution and establishing what success looks like for your business.

Success criteria comprise i) key performance indicators, and ii) hypotheses on their target value, which are closely linked to your business case. Multiple success criteria can be utilised to help achieve your business objectives, provided they are mutually exclusive.

Defining great success criteria means being able to answer these four questions: 

1. Based on the business case, what are the organisation’s strategic priorities with the solution? Should it drive top-line growth, boost productivity, enhance the customer experience? Ensure each business objective of the AI Customer Agent is reflected in at least one success criteria.
   
   
2. What is the scope of the solution? And how are we expecting to build the solution incrementally and expecting value to be realised over several phases? Ensure that the defined success criteria are achievable based on the solution development plan and use case sequence.
   
   
3. Who are the customers and users benefitting from the AI Customer Agents? Anchor success criteria to the right target group, facilitating understanding of user benefits.
   
   
4. What is the existing baseline for each success criteria? Set realistic success targets based on the current performance in each roll-out country.
   

One of the key challenges of defining success criteria is defining the level of ambition. Is success defined by increased sales or by increased customer reach? Do we expect productivity gains, or do we first want to measure user satisfaction? There are two key approaches that can help address such complexity.:

#1 – Think in terms of penetration, adoption, and satisfaction:
It helps to focus on the baseline for penetration, adoption, and satisfaction, and to think that the timespan for expected realisation of the targets typically varies. While organisations might want to measure penetration and satisfaction early on, certain adoption indicators linked to efficiency and top-line gains might be challenging to realise in the short-term. Often, companies expect value realisation after 2-3 months of a Minimum Viable Product (MVP) being live, yet if the solution has not yet been adopted by a statistically significant number of users, the results measured in the period are difficult to conclude on. You need to meticulously be aware of the basis your results are coming from, and how you can use them to push further adoption, penetration, and satisfaction. 

#2 – Tailor success criteria based on your pilot segment:
Never underestimate the segment you are piloting the solution within. The behaviors of your different segments (and geographies) will vary, therefore the ability to prove success needs to be closely linked to the segment and geography you are piloting in. For example, we see that digital adoption in geographies such as the Nordics are much higher than other markets, such as Germany (+23pp. in 2023)³. Deep market understanding is therefore crucial to make an informed decision on your success criteria. 

3) Identify KPIs that measure customer desires and commercial benefits

KPIs are salient for proving value after go-live and optimising the solution for both your customers and your business.

Defining KPIs can be difficult as it involves collaboration between AI and experience experts, strategy, and business development profiles to ensure a cohesive and consistent approach. Anchoring KPIs to the future customer journey where you have mapped use cases of your AI Customer Agent helps both ensure full alignment with the to-be experience and capture every potential value moment generated for the customer.  

As shown in the illustration, there are three lenses to value: 1) customer desirability, 2) technical feasibility, and 3) commercial viability. Brainstorming KPIs across these three lenses helps ensure an exhaustive approach.

Other than being connected to a specific value lens, KPIs must be linked to success criteria and business objectives in the Value Blueprint.

For example, “Share of customers who feel in control of the chat flow“ (KPI) is linked upwards to “Increased customer satisfaction” (success criteria) which is linked to the business objective of “Increased early funnel engagement”.

Similarly, KPIs should logically be linked to use cases.

The KPI of “Share of customers who feel in control of the chat flow” is tied to a “Chatbot” use case, which requires the product team to think business requirements, functional and technical features in terms of a specific target experience for the customer. 

Value realisation is the most critical component of your work

With everyone investing in exploring and scaling GenAI use cases, leaders are demanding tangible value and financial results now⁴.

AI Customer Agents are no exception.

Getting AI Customer Agents right is not just about building a solution – it is about driving measurable value aligned with your business and customer objectives.

Through a holistic, iterative, and structured approach to value, linking business case and business objectives to use cases, success criteria, and KPIs across the customer journey, organisations can set the scene for the expected value framed in a compelling value narrative, laying the foundation for sustained growth and innovation of AI Customer Agents.

Did we get your interest? Reach out to Laura Ivanoff Olsen for a demo of how we have done it in practice.

Clear success criteria and key performance indicators are essential to define what success looks like for the business. However, business success can only be achieved through customer success and customer-centric AI Customer Agents. 
Click here to explore our Customer AI Agents: The Series for more tought leadership on how to succeed with AI Agents.

<sup>Sources: 
1. Autonomous generative AI Agents: Under development, Deloitte, 2024 
2. Deloitte’s State of Generative AI in the Enterprise Quarter four report | Nordic cut, Deloitte, 2025
3. Eurostat Database “Individuals’ level of digital skills”, latest update in 2024
4. Deloitte’s State of Generative AI in the Enterprise Quarter four report, Deloitte, 2025</sup>