I built the BA Centre of Excellence at Equitable Life in 2016. It is still running. That is not something I say to be impressive. I say it because most CoEs are not still running four years after the person who built them has left.
The reason most BA CoEs dissolve is that they are built as programs rather than functions. A program has a sponsor, a budget cycle, and a shelf life. A function is just how things work. The goal when building a CoE is to make it a function before anyone remembers it used to be a program.
Every BA CoE I have seen fail started with a framework. Someone brought in a methodology, trained the team on it, built templates around it, and called that a centre of excellence. Eighteen months later the methodology was on a shared drive that no one opened.
A CoE has to start with a diagnosis. What is actually broken? Where is delivery inconsistent? Where are stakeholders frustrated? Where are BAs reinventing wheels that have already been invented?
At Equitable Life, the diagnosis was clear: no consistent intake process, no shared approach to elicitation, no mechanism for BAs to learn from each other, and no way for the business to understand what it was getting when it requested a BA. Those were the four problems the CoE was built to solve. Everything else was secondary.
The CoE needs three things to stay alive: a standards body, an onboarding path, and a community mechanism.
The standards body is the group responsible for deciding what good looks like. It reviews tools, approves templates, and updates practices when they stop working. It does not need to meet weekly. It needs to meet regularly enough that standards do not drift.
The onboarding path ensures that anyone who joins the BA practice arrives with the same foundation. Not a three-day course. A structured, supervised first engagement where a new BA works alongside a more experienced one and is coached on how the practice operates, not just what to deliver.
The community mechanism is the hardest to build and the most important to sustain. It is the reason people stay engaged with the practice when they do not have to. At Equitable Life this was a monthly peer session where BAs brought a problem they were working through and the group worked on it together. No agenda, no deliverable, no formal structure. Just practitioners talking to practitioners.
They announce the CoE before it exists.
I have seen this happen repeatedly. The CoE gets named, gets a mandate document, gets a communication to the business, and then spends the next six months trying to figure out what it actually does. The announcement creates expectations the CoE is not yet equipped to meet. When it fails to meet them, credibility is gone before the CoE ever had a chance to build it.
Build the CoE for three months before you name it. Let the work demonstrate what the function is before you tell anyone what to call it.
At Equitable Life, the intake model, the peer community, and the onboarding approach were all running for four months before the CoE was formally named. By the time it was announced, it was already producing results. That sequencing made every subsequent conversation easier.
Leadership always loses interest. This is not a failure of leadership. It is the normal pattern. The CoE gets executive attention when it is new and when it is in trouble. The goal is to make sure it is neither for as long as possible.
The way to maintain relevance without requiring constant executive attention is to produce something visible on a regular cadence. Not a report. Something the business actually uses. At Equitable Life, the CoE produced a quarterly analysis of intake patterns: which types of requests were coming in, how long they were taking, where bottlenecks were occurring. That document was used in resource planning. As long as it was used, the CoE had a seat at the table.
A BA CoE that lasts is not a program that was well executed. It is a function that became invisible because it works. The goal is to build something so embedded in how the organization operates that removing it would be more disruptive than maintaining it. That is the only kind of CoE that survives the departure of the person who built it.