7 Vendor Selection Failure Modes in BPM
- January 6, 2016
- 2 Comments
Vendor selection for major projects has never been easy – whether they are BPM projects or not. It’s even more difficult in State governments where there are many rules for participating in state contracts and state projects. These rules effectively cause all of the work for the various states to funnel through just a few major vendors and a few local vendors in each state. When failures of vendor selection happen at corporations, everyone is locked into NDA’s, typically, and the information rarely comes to light. As a counterbalance, however, companies generally have a better track record of changing course when the current approach isn’t working.
When failures of vendor selection happen in the State of Texas, however, most of the records can be obtained and made public by news organizations – and they usually are. Unfortunately this takes a long time (too long to save the failing projects usually), but when it comes to light, it isn’t pretty.
The State of Texas and the Attorney General’s office is currently reeling from picking the wrong horses with their T2 project (which had the potential to have a significant BPM component to it). The coverage in the Austin-American Statesman and elsewhere is pretty tough on the primary vendors involved. The first article I had seen on the subject was this one:
In fact, there is little to show for the more than $200 million spent so far. The American-Statesman reviewed six years of technical documents that tracked the project’s failures and found:
- Accenture has repeatedly failed to deliver on its end of the bargain, with little consequence. In fact, Texas officials have rewarded the company with more money and more autonomy to run the project.
- The project has quietly become a massive offshore outsourcing endeavor. More than 165 workers in India have access to state data and are working on code remotely, despite warnings from software experts who monitored the project.
- Monthly reports showed the project was riddled with complications and disappointing results. Accenture didn’t provide enough staff to work alongside state programmers, those records show, and those the company did provide often lacked the skills necessary to do the job. Often, project monitors noted that Accenture representatives didn’t attend meetings.
- Concerns about the quality of the work by Accenture were raised as early as 2011 and have persisted since.
- When problems mounted and the project fell behind schedule, state officials moved the deadline back. They called it a “rebaseline.” The project has been rebaselined at least five times.
And lest you think Accenture was the only vendor vilified, see what they had to say about Deloitte in a subsequent article covering a legislative hearing on the subject. A $1.8M requirements project turned into a $46M boondoggle.
An assessment of a project can’t get much worse than this. But if you want to read for yourself, the documents behind these statements have been made available by the journalists behind the article here. And specifically, the six-month report by ARiSE is also available.
So how does the State of Texas end up picking two unqualified vendors, and spending $310M on a project that probably should have cost less than $30M?
- The whole process favors insiders and political connections over raw competency – and even when competency is evaluated it is often the wrong competencies.
- State agencies rarely shift vendors when a project isn’t working. They stick with the same vendor, afraid to make changes. Effectively, they just keep digging the hole deeper.
- There are often artificial barriers between phases – between requirements and implementation for example. These barriers were put into place to prevent certain kinds of vendor conflicts of interest, but it is a classic case of the cure being worse than the disease. Requiring a vendor switch after requirements guarantees the first vendor doesn’t have to design for success, and the second vendor has to start from scratch in learning the context. Neither of those is cost effective for our government to fund.
- State agencies, who should have a mission to create jobs at home whenever possible, often outsource the work overseas. I’ll never understand how, in the midst of a recession, state agencies signed off on having their vendor send most of the work offshore, out of country, to contractors that couldn’t easily be monitored for skill, capability, performance, or on-task-focus – rather than keep the work at home. Worse yet, Austin is home to more BPM experts than almost any other city in the US… but their vendor sent most of the work offshore. Our state leadership had an opportunity to create jobs in Austin, TX and instead they chose to create jobs elsewhere, which runs counter to all of their political positioning and reputation – not to mention that it runs counter to the economic interests of their state and constituents.
- Vendor evaluations generally prefer traits that aren’t relevant. For example, when tackling a BPM project with BPM software, around something like unemployment insurance benefits, most government projects will prefer a vendor with no BPM nor BPMS experience, and pick one with experience on the old green screen systems we’re attempting to replace, or experience with the Natural language, or experience with unemployment insurance benefits. None of those things will enable that vendor to successfully lead, implement, and deploy a BPM program.
- Vendor evaluations generally fail to discriminate between a company having experience doing something (for example, Accenture or Deloitte doing work in a different state), versus the specific team members or leads having experience with it. It starts with getting the wrong kind of experience (point one), but ends with not even getting the experience expected, because companies don’t know things, only individuals know things. At BP3 we overcome this by being laser focused on business process (or BPM) as a guiding principle in our work. And by organizing our company to reinforce all of our team members with our collective organizational knowledge. But not every company has this focus nor makes these investments.
- The experience doesn’t go deep or broad enough. A typical general consulting firm will staff the first 20 people on a 100 person project with one or two experts in the subject. And everyone else won’t have meaningful relevant experience in BPM, but they will fit the rate card. When you’re engaging a vendor, find out what the first 10-20 people staffed will look like, from an expertise point of view. If more than 1-2 people are still learning BPM, you should be evaluating other vendors.
Of course, there are other vendor selection failure modes that likely don’t apply in this case:
- Not checking vendor references, or assuming that all the vendors must have similarly good references so not asking for them
- Assuming vend
ors have a similar track record of successful production deployments as a percentage of signed contracts, without checking.
- Assuming vendors have a similar market penetration without checking the facts
- Accepting vendor claims at face value without sanity checking those claims with empirical or statistical data.
- Picking the vendor that is already “approved” for work, rather than the vendor that has the best odds of delivering successfully.
Back to the specific case mentioned above, the question asked is: “Did Texas get it’s money’s worth from this project?”
Company officials knew it was a big job, but they thought they could handle it, said Ben Foster, a managing director with Accenture. Miles asked him if he thought taxpayers had gotten their money’s worth from Accenture.
“I will tell you candidly – not yet,” Foster answered
Okay. We’ll take that as a no. But will Texas take a new path forward? Switch vendors? Try something different to get the original promised benefit of modernization? The answer is no (Emphasis added):
It will take an additional $20 million for Accenture to finish the job, Roy said. The agency has decided to keep the company on T2 because “Accenture will get it built the soonest and at the least cost to the taxpayers compared to the alternative,” he said.
T2 is now scheduled to be launched in July 2018. In recent months, the attorney general’s office has changed its oversight methods, streamlined its bureaucracy and cut the project staff by about one-quarter. Roy said the agency plans to hold Accenture accountable and will financially punish the company if it doesn’t meet the deadline.
This has been the play of the big consulting firms in projects like this all along. You don’t have to outrun the bear, you just have to be faster than the other guy. So the consulting firm’s goal doesn’t have to be success or failure, just creating the perception that any other alternative would be riskier (not necessarily more expensive, just riskier) than staying the course with the current firm. Project success is secondary to making money.
And the State has to look in the mirror on this one. They had many opportunities to look in the mirror and make a decision, with courage, to change the program. And even now, in front of committees in the state legislature, the State is going to stick with the same plan and hope it will get better.
After spending $310M for software that doesn’t work, they’re going to continue to double-down in the hopes that the software will work in the future. But nothing has changed about the team’s basic capability to deliver a solution, nor the complexity of the requirements.