One of the most expensive mistakes in a performance turnaround is declaring it done before it is done.
I learned this the hard way at HCTec.
A large university hospital system had negotiated a favorable contract. To get the pricing they wanted, they accepted a reduced service level: 70% of calls answered within one minute, compared to the 95% within 30 seconds that is standard for healthcare IT support. Their procurement team was satisfied. The numbers were on paper. Everyone signed.
Day one, the users started calling. It was an immediate disaster.
The metric was fine. We were hitting 70% within one minute, exactly as agreed. What the SLA did not capture was what happens to the other 30%. In the model they had negotiated, roughly 10% of callers were waiting five minutes or more. In a hospital IT support environment where a call can involve a system access issue blocking a nurse from a patient record, five minutes is not an inconvenience. It has clinical implications.
The hospital leadership came to me and said we needed to fix it. I told them something uncomfortable: I had staffed to the model they negotiated. I did not have the headcount to deliver what their users actually needed.
What I should have done earlier was push back harder during the negotiation. I should have explained, clearly and without softening it, what 70% within one minute would feel like on the receiving end. I should have protected them from their own decision.
The metric was met. The performance was not. Those are not the same thing.
The problem with KPIs in a turnaround is that they measure what someone, at some earlier point, thought was important.
Contractual measures, internal performance guarantees, SLA thresholds: these represent someone's best thinking at the time of signing. They are a proxy for performance. They are not performance itself.
And this is where most turnarounds get declared done prematurely. The metrics look good. Leadership pressure subsides. The team is exhausted and a declaration of completion feels like a reward for the work.
That is exactly the wrong time to declare it done.
The standard I use is this: a turnaround is complete when the organization achieves 90% of its key performance indicators for 90 consecutive days.
One month of strong metrics is luck. Two months is a coincidence. Three consecutive months establishes a pattern. It shows that the performance is structural, not circumstantial. It shows an operation that is stable and under control, not one that is sprinting through an unsustainable effort.
90% is the threshold because in most complex operating environments, hitting 90% of your core metrics consistently means the fundamentals are working. 100% is a fiction. Any organization that claims 100% on a sustained basis should be audited, not celebrated. 90%, held consistently, tells you the machine is functioning.
When I say key metrics, I do not mean every number in your dashboard. I mean the four or five measures that genuinely indicate operational health. The ones that, if they are all green for 90 consecutive days, you can say with confidence the model is working. The metrics that matter to your customers, your frontline, and your financial results simultaneously.
When you hit 90 for 90, you stop playing defense and you start playing offense.
Defense in a turnaround looks like this: you are reactive. You are managing fires, reviewing reports after the fact, responding to escalations, and hoping the next 24 hours do not produce a surprise.
After 90 for 90, the posture changes. You shift from reactive management to proactive management. You stop relying on reports to tell you what happened yesterday and start building real-time awareness at the point where the work actually happens.
This is when you invest in your frontline supervisors. This is when you build the early warning infrastructure.
A piece of advice I give without hesitation: do not give your frontline managers an office. The moment a frontline supervisor has an office, they believe that is where they are supposed to be. It is not. They are supposed to be on the floor. Coaching. Listening. Catching trends in real time. An office becomes a hiding place. Do not give them one.
Build specific reporting rhythms where the message "nothing wrong today" is as important as "here is the issue I am seeing." Silence in a post-turnaround environment does not mean stability. Your supervisors need to be actively confirming the baseline, not just escalating exceptions. When three or more customers call in with the same problem in the same day, that is not a coincidence. That is a signal, and your frontline leadership needs to be close enough to the work to catch it before it becomes an escalation.
This is when you build the systems that allow the operation to sustain itself. Training infrastructure, coaching frameworks, real-time communication, proactive trend identification. The things that, once built, let the operation run without you managing it every hour of every day.
Turnaround mode is managed triage. Steady state is managed optimization. They are different jobs.
The 90 for 90 standard is the line between them.
Most leaders cross that line too early, driven by pressure from above or exhaustion from below. They declare victory at 60 days because the numbers look good enough, and then they find out that what looked like stability was actually compliance. And compliance decays.
When you stop declaring victory, you start earning it.
Previously in the series · Culture Is Not What's On the Mission Statement. It's What People Trade.