Blog Post

In software, the build-versus-buy debate flares anew

Stay on Top of Enterprise Technology Trends

Get updates impacting your industry from our GigaOm Research Community
Join the Community!

With all the ready-to-rent, easy-peasy SaaS software products on the market, the conventional wisdom has become that you’d be nuts to build your own software. For one thing, companies can sign up and pay for SaaS out of operating, not capital budgets, which has accounting advantages. For another — hey, it’s easy, and face it, software developers don’t come cheap.

But a couple interesting cases cropped up this week, with two big companies bringing more software development and related work back in-house. First, CIO Journal reported (registration required) that Facebook(s fb) is building its own sales software internally, according to Facebook CIO Tim Campos. It will tie into Facebook’s existing crm) account, but Facebook will run the system.

And on Thursday, The Wall Street Journal reported (registration required) that Delta Airlines(s dal) is buying data and intellectual property from Travelport LP so that it can control and manage — and presumably enhance and modify as needed — its own operational and reservations system, according to CEO Richard Anderson.

The argument for buying

At MIT’s CIO Symposium this week, ThomasNet CEO Mark Holst-Knudsen said the biggest mistake of his career was when he approved the building of some software when the company should have bought it. His takeaway, as reported earlier:

“‘You shouldn’t build anything that’s available off the shelf that’s not the source of competitive advantage,’ he noted. For ThomasNet, that advantage lies in its system for managing product and CAD information, which enables it to build product catalogs with specifications that can then be transmitted to distributors with different ERP requirements.”

It’s the definition of “core competitive advantage” that gives big companies wiggle room here, and some will doubtless hear the siren call of vendors who say it’s time for big companies to re-invent themselves through bold new software products.

No one could argue that Facebook’s core expertise lies in developing massively scaled infrastructure and the software that runs atop them, but the decision still flies in the face of convention that holds companies should buy what they can and reserve their BYO chops for the most core, valuable assets. As for Delta, I guess the reservations system is the epitome of core. But again, hiring on software expertise is a very expensive proposition in this day and age.

Little girl shopping with grocery list

ROI is a tricky calculation

When it comes to Return on Investment (ROI) calculations on buy versus build, anything can happen. For one thing, all that SaaS software isn’t exactly cheap — and most SaaS vendors require at least a one-year buy in (and push for three years). For 100 users,’s Enterprise edition lists for $150,000 per year. Many SaaS purchases are eerily reminiscent of the old enterprise software sales cycle that the SaaS players said they wanted to disrupt.

The IT executive of a large Bay Area company, who requested anonymity because he is not authorized to speak on buying decisions, told me his company did the math and concluded that creating its own cloud infrastructure out of open-source software was an attractive way to minimize VMware(s vmw) licensing costs. And it gives the company a way to build exactly what it needs without being constrained by vendor strictures.

When it comes to applications, he added, it’s not always a given that off-the-shelf packaged software is less pricey than building your own, but that has to be considered on a case-by-case basis.

The siren call of new and shiny apps

The ability to build and field truly new applications is a benefit that Pivotal CEO Paul Maritz has been hammering on for the last year. He paints Pivotal’s big data platform as the foundation for next-generation applications that could open up new revenue streams for companies. Speaking at Structure Data in March, he said rather ominously that for some companies in hyper-competitive markets, there is no option.

 “Some players…have clear enough line of sight to competitive opportunities that they’re willing to build afresh, leave IT legacy behind and build a new platform,” he said. “There’s nothing like the gallows to focus the mind.”

The demand for new, built-from-scratch applications will be one topic on tap for discussion with Scott Yara, president and head of products at Pivotal, at Structure next month in San Francisco.

6 Responses to “In software, the build-versus-buy debate flares anew”

  1. Timothy Campos

    I enjoyed reading your article. A bit more on what goes into the decision to build our own software at Facebook. I take the build vs buy decision very seriously. Anything we build will have a maintenance cost in the future that has to be considered. Moreover, when you begin a project, the software that you are “going to build” always looks better than the software someone else already has because you haven’t yet run into the limitations that inevitably show up in software engineering. As such, we will buy wherever we can.

    At the same time, the productivity of our workforce is ESSENTIAL. We look for every opportunity to get efficiency out of our processes. As an example of how much we care about this – we had a project to shave 500ms off the time it takes to walk through our security turnstiles in our lobbies. Slow turn stiles = higher alarm rates, lower security, more staff needed to staff our lobbies, and annoyance for our employees. While many companies would say, “just walk slower”, we felt walking into the office every day was a scaled enough problem that we should invest the time on this. Fortunately, the technology of our turnstiles could be modified to achieve this objective. But far too often with enterprise software, that isn’t easy or possible. Separating the front-end from the backend helps – as we have with many of our enterprise systems, but sometimes to make the adjustments to make a process efficient, enterprise software just gets in the way. Sometimes software defined for the many creates too many tradeoffs for each individual case

    I’ve worked in environments where the arguments about employee productivity are discounted. And it is easy to ignore the impact to employee productivity for individual projects and systems. But the cumulative effect is material and measurable. Moreover, applied smartly, the return on investment is significant. Facebook’s Corporate IT spend as a % of revenue compares very favorably to other companies, yet our workforce productivity is some of the best in the industry. I believe that a firm commitment to ensuring workforce productivity is essential for high performance knowledge based organizations – and a big part of this is the willingness to customize and build software where it can have a material difference for employees and business functions. Thanks for allowing me to share this perspective.

    Timothy Campos
    CIO, Facebook

  2. The key part of the quote “You shouldn’t build anything that’s available off the shelf that’s not the source of competitive advantage” is the second bit: “that’s not the source of competitive advantage”. This means a lot of things are perfect for SaaS: payroll, accounting, email. Things start to blur with more system applications e.g. most people should never build their own database, but if it’s a key part of your application then maybe you can optimise some very specific use case?

    With Facebook you might say that their core competitive advantage is building a scalable website but where does all their revenue come from? Selling ads. Selling. This means if they can get some big advantage from their CRM, that could be important. What if they could mine the profiles of the people they’re selling to?! Or find new prospects – outbound sales to organisations that are getting a lot of clickthroughs but aren’t already advertising? That’s probably not something they’d want to release publicly, so Salesforce isn’t enough.

    What about an airline getting scheduling and ticketing advantages from building their own system? Even a slightly reduced cost in ticketing can be important for such a competitive industry.

    So it comes down to, as you said, how the competitive advantage is defined.

  3. Steve Ardire

    Good piece Barb but regarding your conclusion it’s not just Paul Maritz and Pivotal doing this ;-) The fast emerging Hadoop Big / Smart Data ecosystems and Cognitive Computing Platforms will be best foundation for next-generation applications that will open up new revenue streams for companies.