Business models for linked data and web 3.0

In January, I wrote a post on 7 business models for linked data (with an 8th model added shortly thereafter). Although far from comprehensive, it attempted to illustrate the range of direct revenue vs. indirect revenue models that could justify development of linked data initiatives.

A number of people from the linked data community contributed feedback and suggestions of additional models, including Leigh Dodds (Thoughts on Linked Data Business Models), Paul Groth (Another 5 Linked Data Business Models), Eric Hellman (8 One-Way Business Models for Linked Data), and John S. Erickson (The Evolution of Linked Data Business Models).

However, some of the best feedback came offline from Rachel Lovinger at Razorfish (and author of the blog Meaningful Data). Over a series of discussions, we developed an expanded model:

There are three dimensions in this model, which we phrased as questions:

1. How direct is our revenue? This is the Y axis of the chart. At the top are ways of directly being paid for data, such as licensing and subscriptions. Toward the bottom are more indirect revenue sources, such as using data to drive traffic to a web site or build one’s brand and reputation.

2. Who do we provide our service to? Along the X axis are three categories that identify, from left to right, the openness and accessibility of the service.

On the far left are private partner relationships, where data is provided in closed relationships with partners who use it internally or incorporate it into other offerings downstream. In the middle, data is provided via APIs to a more open developer community, who in turn leverage the data in more public facing applications. And on the far right are web sites that regular users visit or subscribe to, without requiring any technical expertise on their part.

3. Who do we get revenue from? For any given model, the direct user of the service and the source of revenue are not necessarily the same. For instance, with a subscription model the users and the revenue source are the same; with an advertising model, they’re usually two different constituencies.

We used colors to distinguish four classes of revenue sources: blue from direct customers, purple from partner relationships, green from advertisers, and red from “subsidized” sources such as government or non-profit mandates.

In this landscape, we identified 15 business models that offer a good representation of the different ways in which organizations can monetize — directly or indirectly — data publishing initiatives:

1. Subsidized/public service: funded by a government, an NGO, or a regulatory mandate — revenue = funding.
2. Licensing: charge fees to let developers use data in other environments.
3. Microtransactions: on-demand payments for individual queries or data sets.
4. Subscriptions: charge for access to data for a period of time (may have tiers for different levels of access).
5. Freemium: free but limited access to data to sample, but charge for extended premium access.
6. Paid inclusion: charge to be included in the data set or attributed valuable meta-data (what I formerly called an “authority” business model).
7. Sponsorships: charge a small number of advertisers for brand visibility of sponsoring the data.
8. Advertising: charge for ads placed around data on web pages (may also tap into ad networks).
9. Marketplace: provide data to a partner service in exchange for an opportunistic royalty.
10. Affiliate program: provide data streams to affiliates who distribute them in other applications in exchange for commissions on related sales.
11. Affiliate participation: as an affiliate of other companies, combine affiliate product links with data to earn commissions on related sales.
12. Value-add/loss leader: incorporate free or bonus data as an enhanced feature to win customers for another product or service.
13. Traffic generation/SEO: publish data to earn favorable positions in search engines and other directories to generate more traffic.
14. Branding: provide data free of charge on a friendly web site to build brand (i.e., self-sponsorship).
15. Data branding: provide data free of charge to build brand, but it’s the data itself — not the visible manifestation of it — that is the vehicle for meme distribution.

To be certain, this isn’t an exhaustive list. I’m sure there are plenty of innovative business models that are more unique, and that many real organizations will adopt hybrid approaches. However, we do hope this helps people better visualize the landscape of data delivery business models, and that it helps move the discussion forward.

Let us know what you think!

If you’re interested in this topic, I also suggest that you visit the Business Of Linked Data (BOLD) Google group organized by Kingsley Idehen, read the book The Power of Pull by David Siegel, and consider attending this year’s Semantic Technology Conference in San Francisco in late June.