At the same time, almost everyone already understands the need to embrace AI to avoid being left behind. Even though the AI train is already in full swing, there is still time to jump on board.

Even though the AI train is already in full swing, there’s still time to hop on board.

In this blog post, I’ll go through five steps to get you started with AI experimentation, for example by setting up an AI agent. And if you’ve never heard of AI agents before, you can read more about them here.

1. Get people’s approval

The introduction of AI is not just a technical exercise, but like any other business development: a project that aims to advance the work that people do.

The first and perhaps most important step is to ensure that both management and employees are committed to the project. Management needs to understand that investing in AI is business development like any other investment. It may not deliver immediate results, but it can provide significant competitive advantage in the long term.

Employee acceptance can often be even easier to achieve. When they see in concrete terms how AI can reduce manual work or improve workflow, they quickly understand the benefits. It’s important to involve staff from the very beginning – not just when the solution is ready to be deployed. When employees are involved in thinking about where AI could be used, engagement and insights are created.

At the same time, it is important to manage expectations. AI is already working very well in many applications, but in others its capabilities are not quite ready yet. However, the pace of progress is so fast that in just a few months the situation could change significantly.

2. Select the first use-case

AI is not the answer to everything, and its adoption should not be justified by the technology itself. Instead, you should start with a business focus: what are your business goals? What are the business goals? Where in the process is the most friction, errors or time wasted?

A good first step is to choose a limited, clear use case where AI can bring rapid and measurable benefits. Employees are key to this. They know best which tasks are repetitive, manual or frustrating. In our experience, a half-day workshop, for example, where employees work in groups to brainstorm suitable uses for AI and present them to management, can yield a surprising amount of ideas and insights.

Employees are key to this. They know best which steps are repetitive, manual or frustrating.

The most important thing at this stage is to understand that you don’t – and shouldn’t – try to solve everything at once. The pilot should be chosen to be simple enough to implement, but at the same time provide tangible benefits and lessons for the future.

3. Ensure data availability

AI needs data to work, and it is often the data issues that are the most difficult part of the first project. You need to understand what data the AI needs to work, what data the company already has available and in what format it is. Often the data is scattered across different systems or, at worst, stored on employees’ own computers in different documents.

It is often the data issues that are the most difficult part of the first project.

For example, if the goal is to develop an AI solution for customer prospecting, the first step is to understand the criteria used to prospect customers. Then you need to find out where to find the data to do this. In the case of future customers, this data is often not held by the company itself, but has to be imported from external systems.

It is important to figure out how to gather the necessary internal and external data in one place and process it in a format that can be used by AI. Once this step is in place, you are often more than halfway through the project.

4. Put artificial intelligence to work

With the data in place and the first use case selected, it’s time to deploy AI. Surprisingly, this step is often the technically easiest part of the whole process. First, you specify what you want the AI to do in concrete terms and identify any risks or limitations.

The first version of the solution is then built and tested in a small, limited environment. Based on feedback, the model is refined and finalised before going into production. It is important to ensure that the production phase systematically collects experience and findings that can be used for further development.

5. Extend your use

After the initial deployment, the company will have valuable experience of how AI works in practice. This is a good time to look around and consider where else the same principles could be applied. Automating side functions or parallel processes is often quicker and easier when the infrastructure and skills are already in place.

AI adoption is often a self-reinforcing cycle: successful experiments encourage new openings, and as the benefits grow, so does organisational commitment and understanding.

The most important thing is to start

The AI train is no longer leaving – it’s already in full swing. The question remains: will you jump on board now, or will you wait until your competitors are already several stations ahead? There is no perfect moment, but it’s worth taking the first step as soon as possible.

There is no perfect moment (to start using AI), but the first step should be taken as soon as possible.

The important thing is not to get it right on the first try, but to get going. Every experiment, every success and every lesson learned moves the organisation forward. And the sooner the lessons start to accumulate, the better prepared your business will be to realise the full potential of AI.

If you’d like to talk more about starting to use AI in your organisation, book an appointment on my calendar here and let’s talk more!

Shopkeepers and department managers have a major influence on the sale of grocery products: what products are included in the range, where they are placed on the shelves and what promotions are run in the store. For example, there are more than 1200 Kesko grocery stores in Finland alone. Add to this the other chains and you quickly get to the heart of the FMCG sales manager’s headache: how do you manage this whole thing?

The old sales wisdom of quantity-direction-quality also applies to grocery sales.

The old sales wisdom of quantity-direction-quality also applies to grocery sales. If you can’t increase quantity (at least not without the risk of your driver’s licence ending up in police custody for a long time), you have to focus on direction and quality. Fortunately, there is huge potential for increasing sales in these areas.

Direction: planning the seller’s time and routes based on data

Traditionally, field traders have used familiar routes that have been driven so faithfully that you can drive through the route by following the ruts in the road. An attempt can be made to influence the time use of vendors by setting targets for appointments and dividing them, for example, by visiting x number of supermarkets and y number of smaller shops over a period.

This is not yet smart use of time. Sales visits and their content are driven by products rather than by profit potential. However, data is available from companies such as Kesko and S Group, and this data tells us what is sold in which store. And perhaps more importantly, what is not being sold.

Planning is then based on where the stores have the greatest potential to increase sales.

Instead of vendors planning their routes based on their old habits, the system can be used to plan routes on a weekly basis, for example. The planning is then based on which stores have the greatest potential to increase sales. If necessary, planning can even be done on a daily basis if situations arise that require a quick response.

What about those small shops where the salesmen never have time to visit?

This can be done, for example, through marketing automation by adding communication channels to the digital channels (emails, SMS, WhatsApp, etc.), in addition to just conversations with the salesperson. In this way, the number of sales activities can be multiplied without hiring new salespeople. This is handy, for example, when launching new products or selling outdated batches. But more on this topic another time (or book an appointment on my calendar and I’ll be happy to tell you more).

Quality: help your trader sell more

Once the salesperson’s itinerary is planned and he or she is sitting in front of a retailer or department manager in the store, what is the content of the meeting? Instead of tasting the merchant with new products and offering nice POS promotions and subsidies, this meeting can also be based on data.

A trader is above all interested in making a profit. With data, the retailer has a clear understanding of, for example, what products are sold in stores of similar size and serving similar demographics. The salesperson can be equipped with much stronger sales arguments based on the data. The shopper also learns to value the salesperson who makes them more money.

The trader also learns to appreciate the seller, who brings him more money.

As meetings move from tasting products to data-driven discussions on how to increase sales, in many cases they can also move to remote meetings, allowing for more meetings.

From product presenter to retailer adviser

The FMCG business is not easy. There are so many shops, and success ultimately depends on getting the products on the shelves. If the seller can show the retailer or department manager that certain products sell well in similar stores, there is no reason for the retailer not to stock the product.

There is plenty of data available, but the big challenge is to make it useful. For example in FInland, how do you get the data from Kesko and S Group into a format that allows salespeople to plan their time and sales meetings?