Where to start when the Board and Management Team asks for AI agents and data-driven management?

Functos blog guest Toni Keskinen , 180ops

In recent years, companies have focused on building large-scale Big Data and bringing their data together in various Data Warehouse and Data Lake solutions. The data is stored prudently and there is a lot of it. But what next?

Data abundance is a challenge because systems are designed for their own purposes and the data is not originally intended to be combined for business management. A lot of data is not the same as valuable data.

How lots of data turns into valuable data?

Turning data into value requires defining purposes, which results in the creation of value for different data points. Some data is more valuable than others, and each purpose requires a data recipe . Which data points paint a picture of the situation and which variables correlate and represent causality in different phenomena?

Usually, a single data source is not enough to form a broader understanding. A phenomenon under consideration, such as customer churn or a decrease in billing, requires analysis of both the data related to it and the associated data that may explain the underlying reasons.

Structuring data is key to developing the value of data.

The data must consist of entities, such as the company's group, subsidiary and location structures, as well as their service usage, customer relationships and changes occurring in the companies themselves on the customer side.

On the company side, these entities should be linked to the company's operations, i.e. offerings, business operations, channels, salespeople, customer service, etc. In this case, the data structure includes both sides of the equation: a) the customer's customer relationship and situation, and b) the company's operations in relation to the customer.

What benefits does structuring data enable?

Structured data enables customers to be used as sensors in the market. Real everyday life becomes understandable and predictable. Phenomena found in the data can be prioritized and market changes can be reacted to more quickly.

Business operations typically follow the Pareto rule (the so-called 80/20 rule), meaning that 80% of the effects come from 20% of the factors. Identifying these big levers is the cornerstone of generating insights and prioritizing management.

For example, neural networks and agent-based AI can provide answers, but these answers do not contain a reliable answer to the questions why this is happening and what I need to do to change the situation .

When we set out at 180ops to create a solution that would solve this challenge of structuring and obtaining meaningful answers, we set ourselves a demanding benchmark. It also fits well with corporate data managers.

Our goal is to produce the following user experience:

1. Correct information

2. For the right person

3. Duly

4. In an understandable form

5. Through the technologies used for decision-making

The right answers are of no use if they are not seen or understood. Therefore, solving the equation also requires investments in making the information understandable and distributing it. The end result is a flow-through technology that cleans, combines, enriches, analyzes and refines data, and distributes the results to the right channels.

Challenges and Enabling the Deployment of AI Agents

LLM (Large Language Model) and neural network-based solutions efficiently produce answers, but the accuracy and stability of the answers in continuous production do not yet meet the reliability needs from a business management perspective.

According to international studies, 90–95% of various experiments fail.

The problem is not so much the agents themselves, but rather the understanding of their actions.

LLM models and AI agents are effective users and applicators of ready-made data, but they are not built to produce basic information. The results should be evidence-based and validable. The Data as a Product ( DaaP ) ideology is a strong guideline for planning activities.

Structuring, combining and enriching information into entities produces basic information, which can be processed and functionalized efficiently using agents.

For agents to understand the information they receive, it must be shared in a native format, preferably using MCP (Model Context Protocol). Anthropic's open MCP standard is rapidly becoming the preferred way to share information with agents. To operate reliably, agents need an infrastructure to support their operations, with data processing and distribution being a key component.

Build or buy RevOps?

Companies face a path: Data, Processes, and Architecture. Successful navigation along this path is not just technical, but based on a very deep understanding of business phenomena.

In order to define the right data and processes, a strong understanding is needed as the basis for the definition.

Understanding is a hypothesis that needs to be further validated, and during the work, new phenomena and observations are realized that may shift priorities during the project.

For these reasons, it is understandable that many projects exceed budgets and schedules are pushed into the future.

The alternative is to utilize companies specializing in the subject and the technology they produce as a service. The advantages include clearly lower risks, reasonable upfront investments, rapid validation and achievement of benefits, and accelerated learning.

Management understanding is important

A successful data strategy combines three elements: clear goals for what you want to achieve with data, sufficient resources for both technology and expertise, and management's own commitment to acting as an example of data-driven decision-making. 180ops' extensive network of experts also includes Functos Oy, which specializes in clarifying data value creation and data responsibility.

"When the board and management team decide to invest in data management and AI agents, they are making a strategic choice that goes much deeper than technology acquisition. Data is not just the responsibility of IT or D&A functions – it is a tool for steering the entire business, requiring systematic leadership and promoting cultural change from the top level," reflects Mikko Merisaari, partner at Functos.

---

180ops delivers deep customer and sales insight

Finnish company 180ops O y offers ready-made, productized technology. Below are a few illustrations of the benefits of this approach.

Customer relationship status

How many active, new, growing, stable, declining and lapsed customers do we have? What is our risk exposure (dependence on a few large customers). What is the average annual billing of customers and what is the penetration of our offering among customers (cross-selling opportunities). This enables us to understand the situation and define the priorities involved (e.g. preventing attrition or accelerating cross-selling).

Figure 1: Customer relationship status

Growth potential

Potential modeling is used to model the potential value of each offering for each industry, i.e., the wallet size and share of the wallet are defined at the offering level for each company. This enables the analysis, identification and prioritization of market growth potential, for example for sales and marketing investments, and for target setting.

The result is a strong basis for, among other things, strategic segmentation. From a resource allocation perspective, it is possible to distinguish between “farming customers” with no growth potential and “strategic customers” with already high annual billing but still >1M€ growth potential. Small customers with high potential or targets for new customer acquisition are also identified in the market. The definition of customer care models and customer care resourcing gains completely new foundations.

Figure 2: Number of customers distributed according to current and potential value

Figure 3: A snapshot of an individual customer to guide the customer manager's daily work

Figure 4: Whitespace analysis of customer cross-selling opportunities

Sales snapshot

What happens in sales? Where are the measures aimed at, with what goal and how is success achieved? Is success clearly more successful in some offerings or segments? How can sales success be maximized?

Figure 5: Sales pipeline and success in up-selling, cross-selling, and new customer acquisition

Figure 6: Ansoff Matrix-inspired view of the success of different offers in terms of transaction speed, average transaction, and conversion

Background about the author

Toni Keskinen, CPO 180ops Oy

Toni Keskinen is a pioneer in customer-centric business development, having published five books and several international Admap/WARC best practice articles. His particular strength is based on researching, identifying, planning and implementing the logic and phenomena of the customer journey since 2004. Transforming data into understanding has always guided his work. Decades of experience laid the foundation for understanding, which is now transformed into technology. He is the CPO, founding partner and chairman of the board of 180ops Oy.

Interested in continuing the conversation with Toni? Contact us here or send a LinkedIn invite.