Data science is the future of digital sales

When directed at the right problems, data science is a powerful tool for scaling sales capabilities. Off-the-shelf SaaS products make it easy for companies to start experimenting in this area.

How can we make data science insightful and profitable?

Well, what do you want to improve? Sales? Marketing? Or do you want to introduce tools that make buying easier? Or make deliveries more efficient and faster?

Data science for sales should serve exactly these areas, and ideally help you to find new opportunities for improving sales too.

In a company setting, data science should serve the success of the company. Its purpose is not to be scientific for the sake of it, or to discover interesting things. The primary purpose of data science is to generate sales.

Data science and data scientists are experts in solving complex problems that involve large data sets, using methods from computer science, statistics, mathematics and information science. Results cannot be obtained if there is no data, and if the data is not collected correctly, there will be problems with the analysis. For example, if we want to understand why people return their purchases, we cannot analyse the issue if the reason for the return is not recorded, or if the categories are too vague.

Treat an investment in data science as you treat any other investment

Even though data science sounds fancy, it can be treated as a regular investment. What are the costs involved? What are the expected returns? What would be needed to get it to work?

For example, if you compare an out-of-the-box recommendation system with a customised system, is the customised system able to produce sales so much higher that the improved performance can cover the investments needed for developing the system? If yes, over what time period?

Ask the right questions, plan the right actions

Data scientists take responsibility for the steps their work requires. But, in order to achieve business results with data-science methodologies, two things should be done. First, you should find the problems worth solving. And second, you should plan the actions once there are some results.

It might turn out that the time from the first lead to the contact is very long, or that it varies by sales agent, country or another dimension. Or it might turn out that the website is a key factor in omni-channel sales, and – seen from that point of view – it is obvious that it is missing some crucial information. Data scientists can answer questions like “how big is this problem?” and “is it worth solving?, and produce the first hypothesis of “why” something is like it is.

The "after analysis" step is about deciding on the next actions. "Now that we have this information, how do we turn it into business benefits?". For example, we may want to know if some products are performing worse than others, so that we could either fix their product descriptions, re-price the products, or stop selling them completely. Or we may want to know which products are sold together, so that we could use the information in our advertising materials and increase the basket size.

The remedies to the problem cannot be thought of before there’s more clarity about the issue in question, but the data science work should always be about the next step of: "what are we going to do with this information?"

At the end of the day, data science should serve the same purpose as the rest of the company: business success.

Data science in a nutshell

The prerequisite for a data science project is having data. The better the data, the better the analysis.

That said, data will never be perfect, so you may as well start the work today and improve the data collection once you run into problems. Data science work and projects typically consist of the following steps:

1. Problem definition
2. Finding data, cleaning data, connecting data sources
3. Exploratory data analysis and visualisation or modelling
4. Result analysis and communication
5. Taking the results into use: developing dashboards, implementing models as part of company products or processes (this step may never take place if the data indicates there is no point in taking something further)

Thinking from a cost-effectiveness point of view, it’s a good idea to automate reports, visualisations and routine actions once you know what is needed. Data science has its best returns on investment when used for more difficult analysis.

This is how data science scales sales

Data scientists typically work with large data sets. As a result, they are able to perform analysis on a scale that would normally take an army of analysts. Analysing the sales performance of products and product areas is relatively easy when you have, say, 50 products. A bit of work with Excel and asking for some opinions, and the reason why that red jacket keeps coming back from the customers becomes clear.

But what if you have tens of thousands or hundreds of thousands of items? A million, perhaps. Fortunately, data science uses tools and methodologies that can handle and analyse large product sets. Which products or services are sold together? Are there typical shopping baskets or orders? Are some products or services performing worse than others? Can you predict this?

Data science methodologies can generate insights in cases where it is extremely hard to go through the products one by one, even if there would be an army of product managers performing the job.

Data scientists can also work on automating sales processes. Sometimes it takes a lot of manual work to prepare an offer for a customer and close the sale. Obviously, this is not scalable. So when sales are capped by a manual operations phase, data scientists can assist by automating operations. Perhaps a loan decision would not need a personal meeting and a sales clerk writing the papers? Or maybe an offer could be made to another company with a tool that helps the sales person with a prediction of a probable selling price?

Get started with out-of-the-box tools

Even though in-company data scientists definitively understand the business of that particular company, there are a number of out-of-the-box tools you can use to get started.

Digital sales can be improved at many points with SaaS based data science tools. For example, in user acquisition that could mean better budget allocation, ad space buying, or content choices. On an ecommerce website, that could mean recommendation engines or personalised content. It doesn’t usually make sense to build these tools from scratch.

However, you might need help from a data scientist or a data engineer to choose the right SaaS tools and ensure that the essence of these tools, data, is of good quality and that the tools live up to expectations.

Creating better business with data science

Although data science projects include many technical and mathematical steps, at the end of the day the practise should serve the same purpose as the rest of the company: business success. Data science can and should be thought of as an investment that also yields returns within a given time frame.

Data science is a tool just like any other technology. You can only reap the benefits if you use it to solve actual business problems, and act on the learnings. So before you dive head-first into data and algorithms, make sure you know what you are trying to solve with data, and what the potential impact on your business can be.

Learn how to thrive at the turning point of digital sales by reading The Digital Sales Transformation Handbook. Discover how digital sales transformation is changing companies, and how your business can leverage this change through organisational development, customer experience, ways-of-working and technology. Featuring interviews with industry experts, such as Marta Dalton (eCommerce Director for Unilever and Coca-Cola previously), Risto Siilasmaa (Founder of F-Secure) and Antti Kleemola (CDO of VR, Finnish Railways).