The Business Model Project was a semantic database that used a novel taxonomy to classify companies according to their business model(s).

Motivation

No generally accepted definition

The term Business model is often used to describe varying components of a business, but without consensus of what it represents. Depending on who is conducting the research, the inputs may be different. Existing studies have used: resources based views, activities, value-chain, etc. The overall result being that the concept is unclear and not fit for empirical use.

There have been several attempts at proposing a unified business model definition, but these have failed to become universal. Many of the proposed structures lacked sufficient detail to be of practical use. Others were completely theoretical and never used to generate a data-set.

Lack of data

There are currently no publicly searchable data-sets of companies organized by business model. Also, it is unknown if there are any such private databases in existence. It is unlikely due to the previously mentioned lack of a broadly applicable definition. Many of the academic studies that have generated a data-set were single use and limited in scope(e.g. e-commerce or manufacturing only) or built using non-public resources.

Quality issues

While information can be found about a significant number of organizations. Much of it needs to vetted. Many sites, including Wikipedia, allowing anonymous editing. This, coupled with a loose supervision structure allows questionable data to be introduced. In one example, a firm named CustomMade was listed in CrunchBase with a $400k seed round without a reference cited. The Wikipedia article about CustomMade was updated to also include this seed round, but uses CrunchBase as the source. Other unwanted behaviour includes submitting incorrect or irrelevant data as a promotional tool.

Distribution issues

Many public sites focus on certain sub-sets of companies which would prevent them from creating a unified data-set. Wikipedia has a notoriety principle that excludes most firms like start-ups, regardless of their more interesting qualities like a unique market approach. Another popular website, CrunchBase, is mostly used for start-ups and contains little to no information about some of the largest companies in the world if they are not technology focused.

Ambition

The overall goal was to create a mutually exclusive, collectively exhaustive (MECE) taxonomy that was structured to facilitate CAT analysis:

Comparative

Attribute

Temporal

Timeline

The project was broken into three main phases: Academic Research, Stakeholder Engagement, and Technology. Below are milestones in reverse chronological order.

Data

Most of the basic data items originated from publicly accessible websites and directories including, but not limited to: EDGAR, CrunchBase, Wikipedia, and individual company websites. When applicable based on content and context, inferences were made to identify entity traits such as industry, profit motive, competition, etc. All of the unique business model attributes, such as number of unique models, model types and iterations, were determined manually.

For any given entity, information was captured and presented in three formats:

Technology

The Business Model Project was built with Semantic Mediawiki at its core. Apache Lucene provided full text search capabilities. Charts were generated using Graphviz. In browser image creation and modification was powered by SVG-edit. Additional FOSS that were utilized can be viewed at Ohloh. It was served using Amazon Web Services. Tools used for data extraction included Beautiful Soup (Python), iMacros, and HTTrack.

Images

Standard View Structured Data View Defined Concepts Semantic Search The Business Model Project Logo

Video

A 1 minute overview of the basic search interfaces. Originally placed on the homepage of the project's website.

Readings

Hat tip to Zotero for streamlining the research collection process.

Contact

Chris Garner

Page Details

The header is set in Alfa Slab One by JM Solé. Body text and headlines are set in Open Slab by Steve Matteson. Both are delivered using Google Web Fonts. Contact icons are Entypo by Daniel Buce and were packaged using Fontello. The sidebar and images utilize the scrollnav and slimbox2 jQuery plugins, respectively. This page is hosted and powered by GitHub through GitHub Pages.