Data in B2B Sales

Martin Lucky recently chatted with Petr Sobotka and Petr Schwank on the SalesBooster podcast.

Here's a condensed, editorially revised version. You can listen to the full episode here:

What do we mean by data?

What we focus on is collecting external data and signals about companies. Data can be static: is the company large or small, what does it produce? Or it can be dynamic: has something changed at the company, did someone acquire it, or did they buy a new car? This data changes over time. And then there's an entire world of internal data that companies capture themselves through interactions of their sales reps, customer relationships, or systematically from the online environment.

Where does company data come from? How do you get it?

Some data exists because it has to. For example, companies are legally required to publish their business results in the commercial register. Not all of them do, but theoretically this obligation exists. These are regulatory data. Another example is the vehicle register at the Ministry of Transport.

Then there's presentational data that companies create deliberately to inform about themselves. For example, entries in directories, on websites, in radio, television, or newspaper advertising.

And finally, there's incidental data — data created as a byproduct. For example, job listings. They have a very specific purpose, but the listings contain data that can be used for many other purposes beyond just being a candidate who wants the company to consider them for hiring.

For instance, we have about 450 customers and we know that some of them operate in the automotive industry. When we discover that automotive is an interesting segment for us, we're able to identify other companies in the sector.

Do companies actually collect and analyze data?

It very much depends on the company. For example, financial institutions work with company data systematically — they have it built into their models and processes. Plus, these are typically large institutions with their own analytics teams and a few dozen to a few hundred sales reps for SME. That's one extreme.

Then there's the other extreme: most small and medium companies don't have any analytics team, and if they collect data, it's very basic information. For example, when a sales rep wants to verify whether a company is interesting as a potential lead, whether it's large, whether it's relevant to what they do. Segmentation typically starts and ends with finding basic information: company size by number of employees, by revenue, activity information from the Statistical Office. Unfortunately, for the vast majority of sales reps, that's not enough. When you're selling control systems for machining tools, it doesn't matter what revenue the company has or how many employees. That can be a basic criterion, but fundamentally you need to know that the company has machining tools.

How many customers do I need to identify common patterns?

It depends on what common patterns you're looking for. If you don't have at least 2,000 customers, it's difficult. But even with fewer customers, something can be done: for example, you won't create an overall market view, but you can also look at related segments.

For instance, we have about 450 customers and we know that some of them operate in the automotive industry. When we discover that automotive is an interesting segment for us, we're able to identify other companies in the sector. It's a combination of common sense and data. If as a sales rep you wanted to find companies that have machining tools, with normal data you'd have no chance of finding out.

With what kind of data do I have that chance?

Often it's unstructured or semi-structured data. For example, company websites, public procurement, public contracts, and similar sources. For example, a company that signed a public contract to repair an injection molding press for the Technical University in Liberec will have something in common with injection molding presses. Then there are job listings. Companies describe the role of the person they're looking for, and the information from these texts can be mined beautifully.

Let me make sure I understand. For example, I'm going to sell equipment for galvanizing plants. How does it work? How will you help me find similar plants?

If you sell to galvanizing plants, you already know a few. We take all the data from our database that we have on these plants and look at what they have in common with certain hypotheses. For example, keywords, phrases, types of employees they're hiring, perhaps types of machinery they have. In the rest of the universe, we then look for these common traits. And we don't just look at the present — we also look over time, for example at the last three years of companies that were looking for galvanizing plant operators. We identify the brands of machines on which galvanizing services are provided.

Where we identify common traits, we try to quantify how important these signals are. If a company has a galvanizing plant but it makes up only one percent of its activity, it won't be that interesting for a supplier. What interests me more is a company that offers these services regularly and does it commercially.

So data collection is investigative work

It has two phases. First, there's data collection — honest data engineering or coding. Once the data is in one place, a range of algorithms can be applied to process the data and look for common patterns, or evaluate whether there's something important.

For instance, from an ERU license, you can read which company has transformers and how large they are. For a company selling spare parts for transformers, that's very useful information. For a sales rep to find out which company has an ERU license, you can do it manually — which is incredibly painful — or it can be done automatically, which is what we do. But building the automation costs money. If you were doing it as a company just for yourself, the investment doesn't make sense.

So this way I can, for example, evaluate whether a prospect or lead meets requirements, rank them by relevance…

We're also able to incorporate a dynamic component. Let me go back to the car example. If I'm a Mercedes dealer, I'm interested in companies that previously bought BMWs. When they'll renew their fleet can be estimated (typically every two to three years). Our analysis shows not only that a certain company has many BMWs, but also that they bought them a year and a half ago, so now might be a good time to approach them and try offering them a Mercedes next time.

So it's also about not being too early or too late. When someone bought ten new BMWs half a year ago, there's no point in us going there to sell Mercedes now. But I could come offer them some premium engine oil.

It depends exactly on what you're selling. If you're a service center or tire shop working with corporate fleets, you don't really care when the company bought the car, but it's important to know when winter season and tire changes are coming. There can be many things that determine when the right time is. That's why we always work with our customers to understand their business and set this up correctly. This then allows us to process the data in a way that their sales and marketing teams can actually work with it and trust it.

Do you ever find something completely different from what you expected at the beginning?

It happens, but not very often. The magic of systematically collecting and processing data isn't necessarily in discovering something that the sales rep or marketer wouldn't have figured out themselves.

So extreme "aha" moments aren't common?

For one client who focuses quite a lot on Bohemia and cares a lot about how many locations a company has, we discovered that in Moravia there's a retail chain called Hruska that they hadn't heard of before, and now it's a huge customer for them. It wasn't that they never thought to look at retail chains. The point is that we helped them systematically cover the entire market, saving marketers and sales reps time, because instead of searching on Google, the machine processed it for them. The fact that something came up that they hadn't thought of before is the cherry on top.