Implementing artificial intelligence into your organization’s business processes
Previously available exclusively to tech giants, artificial intelligence is now making its way into more organizations’ processes and can be used by all businesses to modify customer experiences, meet the changing needs of the market, soothe the pains of employees relying on guesswork – bring real value. It’s not a question anymore whether AI responds to business challenges in the FMCG industry or not. The question is how FMCG can benefit from AI.
When you’re planning to begin your journey with AI, getting to know your “why” is a good place to start. Why did you come up with this idea? Why AI? Why is it good for your business?
These are only seemingly simple questions – in reality, the hype surrounding artificial intelligence blurs the lines between reality and fiction, preventing business leaders from getting a clear picture of what AI can help them achieve and how exactly it can be used in their organizations. And it gets even more confusing when we consider differences between industries and companies themselves – what data they collect, what processes they have, and what challenges they face.
There’s nothing wrong with starting with an “I want AI” mindset, though. The process will require you to figure out why and how you want to do it anyway. Since many of our clients face a similar challenge – they’re not sure how and where to adopt AI to make it successful – both in the FMCG and other industries, we’re sharing our approach towards identifying the room for AI in your product – the exact place where adopting AI will bring the biggest tangible results.
1. Let’s talk business
To start, we’ll need to focus on your business: discuss the existing processes, list pain points, identify goals to be achieved. The first steps in the process include defining your long-term goal and a short-term goal (a smaller milestone that will help you achieve the main goal) along with appropriate metrics.
Example: let’s think about an FMCG company and one of its managers responsible for sales, let’s call her Claire. The FMCG industry is facing a lot of challenges in this extremely competitive market. The long-term goal for the company might be really general: to increase revenue, for instance. And to make our example simple, let’s say that Claire is responsible only for the B2C sales – getting more customers to choose products of her company over the competitors’.
Now, Claire realizes that there are two sides to this story: it’s one thing that the company has to attract new customers to sell more, but the other thing is that just as new ones are coming, the existing ones are leaving, choosing the competitors’ products. And without them leaving, the sales would grow even bigger.
So, Claire sees an opportunity in AI to help increase revenue in the company. She then sits down with her coworkers and data scientists and thinks about how to start. There are many AI-driven tools that can help in the sales process: trends and demand prediction, dynamic pricing, and recommendations, to name a few. They all look promising, but Claire realizes that at this point bringing new customers in is very challenging – and she’s absolutely right since it’s 5 times as expensive to acquire a new customer than retain an existing one. She also sees that the current customer retention strategy is inefficient. Knowing all that, she sees increasing brand loyalty will bring the most value to her company. Learning which goods will be in demand for specific customers at a specific time will improve the retention rate – so it makes sense to start there.
The right start
AI doesn’t start with development, though, The beginning is all about a thorough analysis of your organization’s goals, pain points, and its current state. So before you get to the development part, it’s good to prepare your data strategy. Your data strategy is a document that gathers all the necessary information about your business, goals, information about data, and the details of your AI adoption.
A data strategy ensures that the data collected by the company is actually managed like an asset. Data strategy includes elements of business strategy, goals for the project, data requirements, KPIs. Each data strategy may be different and consist of various elements, adjusted to the organization’s needs.
2. Find the business case
Your data strategy cannot work separately from your business strategy. You want to use data to drive business results, so the first step is to look at the business objectives and priorities. You need to select a use case that will solve a business problem you’re facing, and you should only focus on what’s doable. You don’t have to go for “the one” – the only right use case. Absolutely not! It’s fine to list many use cases that you see fit for your business. Next, when you analyze the available data and calculate the ROI, it will be easier to prioritize these business cases.
3. Identify the goals
What’s the long-term goal for your business? How can your data science project help you achieve it? You probably know it already, and it’s a very important element of your strategy, but you also need quick wins – smaller milestones that will manifest that you are on track. Have a look at the use cases you’ve listed. Which one of these will bring the value you’re looking for the quickest? That’s the one to start with then – and all quick wins are steps towards the main objective.
Getting back to Claire: knowing that the demand prediction and personalized recommendations are the lowest hanging fruits, she may want to:
- run X marketing campaigns based on the demand prediction
- improve CTR of the email campaigns by 3% thanks to personalization
And so on. It’s much easier to plan the project when you know what you want to achieve and when you’re able to put it in some timeline.
4. Know your data
It’s time to answer some questions considering data. Think about what data you have, what data you need. Now, check if the data you have is actually usable.
At this stage, you’ll also have a closer look at data storage, governance, managing the results of the model you want to develop. It’s an important step in the process – don’t forget that the model can only be as good as the data that’s fed into it. So if there is no data, figure out how to get it.
Even if you don’t have detailed sales reports showing when there was a “hype” for specific products – and the seasonality is quite an often case in the FMCG – you can monitor Google Trends and see how searches for these specific products changed over time. If there is data, know what data it is, where it comes from, how it’s organized.
5. Define the skill set
Apart from the technological aspects, you should also consider your team composition. Do you have the skills you need to deliver the project? Do you want to train your staff? Hire an in-house data science team? Do you want to partner with another company? All of these have their pros and cons.
Additionally, you have to think about the staff who will be working with the AI, not on the AI project – the end-users. If you’re implementing AI into your sales processes, how does that change the work of your sales staff? How do they interpret the results, what do they use them for? Maybe you see that other use cases will be more valuable for those people? It’s not about 100% tech knowledge, but also about understanding the value AI brings and the way to work with it. AI should not replace your team. It should help them achieve the best possible performance.
6. Plan the core activities
With the business cases selected, tech and staff requirements analyzed, you can outline the activities that have to be performed during the process. You don’t have to design a very detailed project roadmap, but identifying core activities will go together with identifying the skills and know-how you need in the project.
7. Measure – always measure!
Identify appropriate KPIs to verify whether your project is on track. Check these on a short-term and long-term basis, and adjust if necessary. No strategy can be pursued without KPIs – a strategic approach cannot be lacking information about progress, success or failure, and relevant metrics.
Hint: distinguish business metrics from the scientific ones! This will help you avoid a situation when despite having a “perfectly” working model, you see that AI is not bringing you business value. So where to use these metrics?
Use scientific metrics to evaluate if the AI model is reliable. Can it predict that the specific product will be in higher demand? Does it fit your accuracy margin? If not, tweak the model, or try a different approach to solve this case.
Then, if the results are satisfactory, focus on the business metrics relevant to your FMCG company. Do we have a 20% increase in revenue, using AI? If not, maybe you should change how AI is implemented? How do people use the results? Or something strange is happening on the market, which will impact your normal way of working?
8. Become data-driven
Your staff will have to learn how to work with AI solutions and how to use the insights in their everyday work. You need to make sure it becomes their habit to make data-driven decisions. Data-driven organizations have processes that enable employees to acquire the information they need, but they also have clear rules on data access and governance.
9. Do not go at full throttle (at least from the start)
AI can bring you amazing benefits like significant churn reduction. But when you allow your emotions to guide you, everything can turn around and not work as intended. Therefore, starting smart with AI is also starting small.
When you implement artificial intelligence step-by-step, you minimize the risk of failure and potential negative impact on the business. You also create more opportunities on the go, which might give you amazing results even without full implementation. Think about it. How great would it be to know when one of your biggest clients is considering leaving you? Such knowledge is possible to get and you really don’t need to go through a company-wide AI adoption to have it.
How to find your AI use case
Finding the right place for AI in your business is sometimes not an obvious choice at all. In order to know what that right use case is, you need to think about problems first.
Cassie Kozyrkov, Google’s Chief Decision Scientist, has a cool trick for that:
I want you to imagine for a moment that there is no machine learning. It’s all a hoax; there’s just an island in the middle of the ocean somewhere with a bunch of my friends behind computer screens pretending to be an AI. When you send them an input, they quickly send back a decision.
When you imagine it like that and think that you could have them do something for you, what would that be? You’d probably give them something that takes time, is complicated, repetitive, mundane. That’s the right sort of task AI can handle. Even with the long-term goal as general as “increase revenue”, there will be many ways to achieve it. Be it through adopting AI somewhere in the sales process, automating some jobs, making a product smarter. Getting to the right use case may sometimes be difficult, and that’s fine. That’s why it’s good to use the support of your coworkers (people who know what’s the most annoying thing in their work that they’d like fixed) and data scientists who will hear you out and help you find the answer. If you are interested in a more “in-depth” guide on how to implement AI in your business, check out our ebook on how to start smart with your AI adoption.