Implisense Research

Current project: Customer Prediction Platform (CPP)

Even today, most people still work in B2B marketing with few data points to segment potential customers and address them correctly. Typical data points for segmentation are branch affiliation, turnover or number of employees. However, these dimensions are hardly suitable for assessing unknown companies according to their affinity for their own products and services. This is a major problem, as companies increasingly rely on content marketing to generate inbound leads and have to evaluate these prospects in order to allocate them correctly.

The Customer Prediction Platform (CPP) is a new middleware to be developed for the evaluation of corporate customers on their customer potential. These evaluations are made available to various end applications via an application programming interface (API). The end applications include the systems widely used in B2B marketing, such as those from SAP, Salesforce and Microsoft Dynamics.

Classification of companies according to size classes with CPP

The CPP addresses a number of user problems, because

  • it is increasingly relying on inbound marketing to attract new potential customers, and how difficult it is to evaluate unknown companies with internal data
  • the retrieval of publicly available data on companies is frustrating for the users of CRM systems and can be automated to a large extent as a service
  • it is not a matter of collecting undirected information about corporate customers, since these are usually not evaluable by the end users and quickly lead to an overload
  • the availability of internal data on customers and prospects often does not allow reliable analyses of customer value (e.g. according to ABC analyses) without having to carry out costly data cleansing and enrichment.

Meanwhile, a large amount of public information about companies is generated on the Internet, which can be used for ad hoc evaluation of individual companies by including open data (regional data, sector reports, financial data) to form a new type of data basis. The requirement for this is that technologies are used which can automatically process a structured information basis for automated analysis from previously unstructured data, e.g. from websites, job advertisements or press articles.

In order to implement the CPPs, which are designed to ensure the future growth of Implisense and enable us to position ourselves in international competition, we have three objectives for further technical development:

  1. Analysis: Improvement of methods of forecasting of target customers
  2. Integration: Easier integration into existing CRM solutions
  3. Internationalization: Transferring the solution to other countries and languages

This project was co-financed by the European Regional Development Fund (EFRE).

Completed research projects

European Company Explorer Platform (ECEP)

Project description: The open data project by Implisense funded by ODINE. With ECEP you are able to search for any arbitrary company properties on company websites in Europe. Analyze and compare the latest company trends in different regions and industries. And the best is, you can download the results for free! The beta version is now available for United Kingdom.

Project website:

Project duration: May 2016 – October 2016

Linked Value Chain Data (LUCID)

Project description: In LUCID we research and develop on Linked Data technologies in order to allow partners in supply chains to describe their work, their company and their products for other participants. This allows for building distributed networks of supply chain partners on the Web without a centralized infrastructure. LUCID is funded by the german Federal Ministry of Education and Research (BMBF) in KMU-innovativ: Informations- und Kommunikationstechnologien initiative, which part of the IKT 2020 – Forschung für Innovation funding programme.

Project website:

Project duration: October 2014 – September 2016

Insights into the German corporate landscape

As Data Scientists we are always interested in new, perhaps surprising insights into a company landscape and the behaviour of companies.

From time to time, we process the results of our internal experiments to share them with you. Some of them have been so well received that we have integrated it into the software. Other things are exciting on their own and therefore remain an experiment in the lab.

Where are farms affected by storms?

For an insurance company, it can be interesting to know companies that are affected by storms. This could be helpful, for example, in the sale of insurance policies that cover corresponding losses. As part of a pilot project, we have geolocated 1,300 farms with outdoor facilities and assigned current weather data to them, which indicate wind force. This results in a real-time wind map (see figure).

Windkarte für 1.300 Agrarbetriebe

What were the reasons for outages according to fire brigade reports?

Feuerwehreinsatzkarte 2012

It was tested to bring the operations of fire brigades from Germany in 2012 in relation to the geographically neighboring companies to understand reasons for business disruptions.From this question an interactive application was developed, which provides information about focal points and individual incidents. In addition to the thematic sorting and the location of the inserts, a heat map was used as a calendar that clearly shows when potential business interruptions are to be expected.

Which statistically significant terms are used in a branch?

The illustration uses the Reingold-Tilford algorithm to illustrate the statistically significant terms of companies belonging to an industry. The figure shows only the 20 most important terms from selected industries. Information on how to create these images with the D3.js visualization can be found here. More information on the dynamic industry classification can be found here.

How many events with effective date took place on a daily basis on the 2 million companies listed in the Commercial Register in 2012?

You can call up the daily values of the 500,000 individual values if you hold the mouse pointer over a calendar day. You can also access the embedded page via the following link: Heatmap. If you can’t see a picture, your browser is not HTML5 compatible. Here you can view a statistic version.

The commercial register has just over half a million dates for 2012, when changes relevant to the commercial register will come into effect for registered companies. On 20.12.2012 alone, there will be more than 3,000 per day. From such an overview it can be deduced that many trigger events should occur around March and December. During these periods, you should be prepared for changes in your own customers and partners.

For information on how to create this picture with Mike Bostock’s D3. js, click here.

3.8 million companies in one chart

How are 3.8 million businesses distributed among German districts and independent cities? We’ve interrogated and visualized the data from the Federal and State Statistical Offices:

The above illustration uses D3’s Pack Layout Algorithm for a representation of the 3,810,594 companies (31.12.2009) according to districts and independent cities.It is immediately apparent which degree of concentration is present in the metropolises of the Republic. The data come from the Federal Statistical Office.