Come for the service, stay for the data

Equifax’s best business — $2.6 billion in revenue, 51%+ margins, 6-7% CAGR — started life as a call center handling other people’s HR paperwork. How it got from there to here is one of my favorite B2B business-model stories, and it’s where this post ends up. But first, some history.

Come for the tool, stay for the network

In 2015, a16z investor Chris Dixon penned an influential article titled “Come for the tool, stay for the network.”  He argued that solving the cold-start, or “chicken and egg,” problem of getting the network flywheel started is difficult. A solution he offered is to provide a “single-player mode” tool first to one side of the network to reach critical mass.

As a B2B nerd, the examples Dixon provided at the time did not resonate with me.  One was something called Delicious, which I guess I missed — it died within a couple of years of the post.  The second example was Instagram, which is a great business, but not my cup of tea.

There are, however, substantial examples of the “come for the tool, stay for the network” strategy closer to the B2B world:

• OpenTable was an application for restaurant hosts (single-player mode) before it became a marketplace.
• SAP Ariba and ServiceChannel (both businesses where I worked) employed this strategy successfully, moving from single-player procurement platforms to two-sided networks.
• AvidXchange, and to a lesser extent Bill.com, started as Accounts Payable software and then built payments networks.
• Faire gave independent retailers free inventory/ordering tools, free returns and net-60 terms, which bootstrapped a wholesale marketplace connecting them to brands. 
• The Bloomberg terminal started as a source of bond data, but over time became indispensable to traders because of its chat function.
• Cvent built a marketplace business on top of its event registration business and monetized the hotel side of the network, not just the event-planner side.

Be Careful Transitioning from Tool to Network

Making the transition from tool to network is tricky.  Carta is a cautionary tale of the limits to this strategy.  It started as cap table software and then tried to build an exchange for trading private stock (CartaX).  Customers did not appreciate Carta hawking their private shares, and CartaX was shut down within days after customers complained.

If you are going to transition from single-player mode to network, your initial clients cannot be surprised; you can’t use their proprietary data in the process; and they need to accrue some of the network’s benefits.  (The subject of another post perhaps.)

Come for the service, stay for the data

Over the years, I’ve been interested in a variant of Chris Dixon’s theme:  B2B businesses that started out as single-player service companies and then became shared data companies.  Two transitions: single-player to multi-player and service to data.

Most B2B data businesses started as people-intensive information gatherers who shared data through publications and directories since there were no computers.  When Dun & Bradstreet’s predecessor started with credit reporting in the 1840s, clients went to D&B’s offices to hear their customers’ credit reports read aloud! Eventually, D&B sold print subscriptions to the credit reports.  More than 100 years after its founding, D&B introduced digital products, as well as the D-U-N-S Number.  Similarly, Insurance Services Office (ISO), Verisk’s core business, was a services business, but it too was a consortium model from the start.  

But some B2B data businesses made the transition from a single-player services business to a pooled data business or network much better than Carta did.

Single-Player Mode to Pooled Data Asset

FICO

FICO, the famous credit scoring system that still dominates the consumer market, started in 1956 working with one lender at a time —single–player mode—using each lender’s own data and algorithms to predict customer credit behavior. It was not until 1989 that FICO introduced a generalized score that all three credit bureaus adopted and sold.  (Even later, the three credit bureaus formed a joint venture to create a competitive product, VantageScore, in response to FICO’s outsized share of the profit pool in that industry.)

Nielsen

A.C. Nielsen Sr. started as a consultant doing bespoke projects.  When he hard-pivoted into a data business, he did something really smart.  Instead of asking manufacturers to share their data in a give-to-get model, he went to the customers of his eventual customers — retailers.   First he went to drugstores, then grocers, to physically determine how much product was moving off the shelves.  He then sold this data back to manufacturers.  Along the way, he also coined the term “market share,” and the race was on!  Most CPG and pharma markets have followed a similar path since.

But these are ancient stories; The Work Number is much more recent and represents an entire change in business model to become a data business.

The Work Number

The Work Number is part of Equifax’s Workforce Solutions (EWS) business.  EWS generates roughly $2.6 billion in revenue (2025), grows 6-7% annually, and has 51%+ adjusted EBITDA margins.   It is Equifax’s best business by a wide margin and one of the world’s great data businesses. It also started as a single-player, bespoke service for employers, not a data business!

Equifax acquired TALX in 2007, but TALX spent its first 20 years as on-premise software and a business process outsourcing company with modest margins.  It handled HR paperwork for large enterprises, including unemployment claims, I-9s, and eventually employment verification.  TALX convinced large companies to send them payroll data and then set up call centers to handle the calls these employers received from mortgage companies, prospective employers, and creditors seeking to verify their employees’ employment.  The company’s customer was the employer, and TALX was handling a painful, non-value-adding task.  Classic BPO stuff.

Someone smart at TALX realized the company’s business model could be completely reversed.  Instead of charging employers to manage these call centers, they could digitize the business, build a database, and charge people who needed employment verification a fee each time they accessed it.  TALX locked up the most employment data of any verifier (through first-mover advantage, acquisitions, and revenue-sharing) and built a phenomenal business. It’s my favorite example of a company competing with an entirely new business model.

The Same Caution Applies

Whether it’s a service or a tool, moving from single-player mode to a shared data asset is tricky.  (And it’s only getting trickier as companies seek to realize the value of their proprietary data.)  Forming consortia, paying for data, using only non-core data (like employment verification), and making sure data donors get more value from sharing than they perceive they are giving up may all be necessary.

Single-player services can be kindling for great data businesses.  Just make sure you are in lockstep with your customers on the benefits they will receive from moving to a shared (or network) model before starting.  Ask Carta.