Jakub Vraspír on Data Projects at BizMachine

You know BizMachine as a data powerhouse for B2B salespeople. Did you know that our work often doesn't end with data delivery? Jakub Vraspír, who spent several years in sales and project management before joining BizMachine, describes the services we offer clients to develop and streamline their sales teams. In his year at BizMachine, Jakub has worked on projects in media, e-commerce, telecommunications, and pharmaceuticals.

Client Projects — From Quick Analyses to Multi-Year Partnerships

Our projects build on top of the data we provide to clients. These projects are either long-term (lasting up to several years), short-term (such as monthly diagnostics), or one-off ad-hoc analyses.

In long-term client projects, we don't just deliver data — we try to help with the entire sales process and ensure that salespeople can use the data as effectively as possible. At the same time, we take their feedback, grow together with clients, and learn. When we work with a customer for two or three years, we can see changes in a long-term context and, based on experience and data, elevate the entire partnership to a much higher level.

There are also short-term and one-off projects — these are ad-hoc requests for various analyses, diagnostics, or segmentations based on specific briefs.

In the case of segmentations (or micro-segmentations), we primarily try to combine our experience and data know-how with the client's knowledge of their market. On this basis, we decide which data and analyses to combine to most effectively reach the right group of companies. It's very important here to also think about the salespeople or marketers who will ultimately work with these opportunities and convert the provided leads into real business opportunities. That's why it must be clear to them on what basis we selected the companies and why they are potential customers for their services or products. Thanks to our data, a salesperson knows, for example, that they're about to call a company that owns a manufacturing plant, has Czech owners, 50% higher hiring than last year, is looking for quality controllers for production, and tends to invest in more expensive technologies. They therefore have all the key information and understand why this particular company might be interested in purchasing camera systems that analyze production quality.

We often see companies requiring manual analytical work from their salespeople. Salespeople first browse through CRM, various databases, and the internet to identify potential customers, then search for important information about them, and finally try to close deals. This is largely a waste of a salesperson's time. At BizMachine, we believe it's much better to serve salespeople relevant leads every month or quarter, selected based on quality analysis and market experience, and let them focus on selling. 

Segmentation Models

For clients thinking about (micro)segmenting their market and wanting to make their salespeople's work more efficient — or to find out if there are areas of the market they haven't yet covered — we essentially offer three models: Attractivity scoring, Similarity model, and Potential-sizing regression model based on invoicing. Each model suits a different use case, and we often combine them.

1. Attractivity scoring is a model based on hypotheses and the assumption that the client knows the characteristics of their potential customer. Together, we compile a list of individual signals that a given company should meet. For example, if we want to focus on companies that own an e-shop, the examined parameters might include signals like presence on Heureka or Zbozi.cz, a certain revenue level, positive reviews, and so on. We then assign a score to each signal based on importance and run the model on the given market. Companies with the highest scores should then have the greatest potential. It's a great model for clients entering a market with a new product. It's also very transparent for salespeople and easily adjustable based on field experience.
2. Similarity model looks for similarities. The client identifies a set of "good" customers (from experience or based on invoicing) and our machine learning model identifies other similar companies in the market and quantifies the degree of similarity. This model relies on the client's real data and can lead to discovering interesting signals that we might not have found with Attractivity scoring.
3. Potential-sizing regression model is in many ways the most accurate, provided the client has sufficient invoicing data and share-of-wallet data for a specific product or service. With this model, we can not only model what the "ideal" potential customer looks like, but also quantify how much a given potential customer should spend with our client.

Supporting Sales Teams

Beyond purely data-driven work, projects are increasingly focused on "softer" areas, such as organizing sales teams, properly setting up incentive structures, work systems, and more. We advise clients on how their sales teams should be structured, how to work with tools most effectively, and how to segment salespeople against the market. We typically sit in on their meetings and become part of their sales team. We try to leverage our experience and work with them to improve every part of their sales process, from choosing the right call script for outbound calling, through tool implementation, to setting up new sales channels.

We advise clients on how their sales teams should be structured, how to work with tools most effectively, and how to segment salespeople against the market. We typically sit in on their meetings and become part of their sales team.

‍

‍

‍