A towering decision
For Tower Records, the question was whether to develop or outsource a better search engine
As outsourcing budgets increase and hosted applications proliferate, CTOs are finding that the build-vs.-buy choice is no simple matter. CTOs must examine their options from all angles — expected ROI, in-house development and maintenance skills, status of other projects, upstream buy-in, and more.
“Almost anything we can develop on our own, there’s probably somebody out there somewhere who is also developing it and selling it. You have to lay out your priorities [for the tech decision] and keep them in mind through the entire process,” says Ertell, director of Internet technologies, who oversees the Sacramento, Calif.-based music retailer’s online technology and strategies.
Evaluating resources realistically
For Tower, the need to develop a new site-navigation, browsing, and search tool was sparked by customer demand. They wanted a more dynamic online browsing experience, something akin to the in-store customer experience. “In the physical world, it’s ‘location, location, location,’ but on the Web it’s ‘performance, performance, performance,’ ” Ertell explains.
Tower had been building its own browse applications “pretty much from the genesis of the site,” Ertell says, using an ASP front end on an Informix database. But those applications had been hard-coded and lacked the dynamic browsing capabilities TowerRecords.com customers wanted. “Basically, we built daily maintenance tables that contained the titles that fit our predefined browse options,” he says.
Ertell and three programmers started building the new functions by improving the existing navigation technology. “We could see that it was going to be a long, tedious process to get where we wanted to be and probably expensive in terms of the hardware we were going to need.”
Ertell then reviewed the project priorities. Three must-haves in the browsing/search application qualities emerged: functionality, time-to-market, and price. The decision whether to continue to build in-house or outsource it then rested on what would have to be sacrificed.
“We probably could have eventually gotten the functionality [by developing it in-house] — we have a great team — but the other two things would have been out the door,” Ertell adds. To attempt building the type of browsing application that Tower now uses, Ertell estimates he would have needed at least five dedicated project staffers — a project manager, a DBA, and three programmers — and lots of time away from other projects in progress.
Ertell has recommendations for other CTOs facing the build-vs.-buy decision: Review skill sets of in-house talent, then assess whether the project is worth the team time required or whether it is better to purchase an application or partner with another company to achieve results for similar costs. Even so, Ertell does not recommend outsourcing core functions; in Tower’s case, that was checkout and order processing. “It’s very important that we maintain that and stick close to the customer … I’d never want to give that up,” Ertell vows.
Another important point to consider when faced with a potential development outsourcing decision is staff morale, Ertell says. “Outsourcing too many projects may lead to good workers leaving because they feel like they don’t get to work on the ‘good stuff,’ ” the vice president says.
Surveying the outsourcing options
When the decision to outsource the application was made, Tower began the process of deciding which of several vendor solutions presented the best solution for Tower. Last spring Ertell signed on Cambridge, Mass.-based Endeca and its InFront catalog navigation product, which offered “the functionality we wanted and the performance of the functionality — the fact that [the tool] was very, very quick,” Ertell says.
With Endeca’s technology, visitors to Tower’s U.S. site can now refine a search for music or movies and select the way they want to view results. “In the case of music, you might start out with pop music, then refine that down to rock, then classic rock … then say you want to look at compilations that have Elton John on them, and only ones less than $10, and so on,” Ertell explains. “You’re getting right down to what you’re looking for, even if you’re not really sure what you’re looking for.”
As CTOs address the build-vs.-buy decision, they are exploring how vendor offerings mesh with their current technology and business goals, says David Gourley, CTO of Endeca. “We definitely do talk to engineering teams who are trying to make that decision, ‘Do we build it ourselves, or do we try and buy something from a vendor?’ A lot of folks feel like the needed control over building a proprietary solution — so they can exploit their own domain knowledge — is lacking in a lot of products out there.”
When talking with customers, Gourley sees “a lot of questions around the essence of the technology, how it’s going to fit into the overall environment, and of course that’s important [to any technology decision],” he says. “But there are a lot of folks now trying to focus on the business aspects of how to use and exploit the software to really promote the things you want users to experience when they come to your site.”
But with any build-vs.-buy decision — particularly for proprietary technologies or those that are customer-facing, Gourley says it’s important for CTOs to forecast what will be needed to support the company mission six months out — or even longer. “When new products come out, are you going to expand to new areas? How is that going to affect design decisions you’re making today, and are you willing to invest in a team that’s going to evolve that?” he asks.
In making the final decision on which vendor to choose, Ertell and team conducted test runs that pinpointed where more customization would be needed. After opting to go with Endeca, the Tower team conducted several internal and external trials, starting with small, select customer groups before rolling it out first on their video site and conducting customer focus groups prior to further implementation.
Finding a vendor that would implement changes as the customer feedback came in was important to Ertell. “We wanted to make sure the customers did like this [functionality] before we committed fully,” Ertell explains.
Gourley says these tests were important to Endeca because they provided real-world feedback on how customers use online search and browsing applications. “There’s a lot of cloudiness in the way people are going out and using the technology that’s out there,” he adds. “We had some surprises along the way as to how [Tower’s] users were interacting with their site, and it led us down different paths.”
Making the business case
Ertell says that during the decision-making process, he, Tower Business Analyst Lisa Scovel, and former Vice President of Operations Jon Feidner were involved in bridging the gap between the business and the technology sides. In addition to running the technology and overseeing that process, Ertell says that he “did a lot of coordinating between different points of view. I played devil’s advocate an awful lot to make sure we were making the right decision,” which was critical in making sure all contingencies were considered.
Although build-vs.-buy decisions may seem straightforward after the pros and cons are laid out, Ertell says questions always remain until — and sometimes after — the final choice is made, no matter how smoothly the decision-making process flows.
“[Outsourcing] is a tough philosophical issue for some people, but also you give up a certain degree of control when you outsource stuff, and that was definitely a big consideration,” Ertell explains, adding that in this case it took conducting the trial run on the Tower video site to lay to rest any leftover, lingering concerns about the decision to outsource.
Since rolling out the site-browsing application with Endeca, Ertell says Tower has been getting positive feedback from customers and continues to make customer comments an important piece of their decision-making process.
“At the end of the day, if you have an interesting technology, it can’t just be interesting, that’s not going to drive real results,” says Endeca’s Gourley. “The important thing is to tie that back to what the users are actually trying to accomplish when they come to your site or company.”
