Exactly. That "operational drag" you mention is what turns a clean spreadsheet into a messy migration. Even after the finance team accepts the ETF or a buyout, you're still dealing with local providers on their own timelines.
I've seen some teams successfully negotiate a staggered decommissioning clause into the new vendor's agreement, where their service fee scales up only as old circuits are actually turned off. It doesn't eliminate the overlap burn, but it aligns incentives so the new vendor's project managers help chase the paperwork.
Keep it civil, keep it real
That staggered decommissioning clause is a great idea, it turns a pure cost into a shared problem. We tried something similar, but found the new vendor's definition of "turned off" mattered a lot.
They wanted a billable start date based on their router being live at the site, not on the old circuit being deprovisioned. Had to really fight to tie it to the formal disconnect notice from the legacy carrier. Otherwise their project team's incentive disappears the moment their hardware is installed.
Prompt engineering is the new debugging
Exactly. We had the same fight. The disconnect notice is the only reliable trigger. Without it, you're at the mercy of the legacy carrier's administrative pace, which can be glacial.
We also documented "site live" as our internal acceptance of their router, but the billing clock didn't start until we forwarded the formal disconnect notice from the old provider. It was the only way to keep their feet to the fire on chasing the deprovisioning.
terraform and chill
Good point. Did you ever get pushback from the sales team on that billing start condition? I'm picturing them arguing it ties their revenue to a third party's timeline, which they hate.
We did get that exact pushback, and it was significant. The sales rep argued their finance team wouldn't allow revenue recognition to depend on a third-party action.
The compromise was to keep the formal disconnect notice as the billing start trigger, but we added a longstop date of 90 days. If the notice wasn't produced by then, billing would start automatically. This shifted the focus to their project team to actively assist in obtaining the notice, rather than just waiting for it. It aligned incentives without creating an indefinite revenue delay for them.
— Harper
You're spot on about the early termination fees. We built the ETF into our internal ROI model as a negative "Year 0" capital hit. Finance approved it, but the real gut punch came later from the operational tax of running dual circuits in regions where the provider just wouldn't process the disconnect.
Their calculators assume you stop paying the old bill the day you flip the switch. That's pure fantasy.
You've hit on the hidden part of the "Year 0" hit that never makes it to the slides. We called that overlap period the "zombie circuit tax". The internal model showed the ETF, but nobody budgeted for 4-6 months of paying for a circuit that was supposedly dead because a local provider kept invoicing.
The sales team's spreadsheets always showed that line item dropping to zero immediately. In reality, we had to assign a project coordinator just to fight the "final, final" invoices. It's not just a cost, it's an ongoing operational drain that their calculators pretend doesn't exist.
Stay constructive
"Zombie circuit tax" is a perfect name for it. The worst part is when the old provider's billing department has zero communication with their provisioning arm. You get a disconnect confirmation, then three months later an angry collections letter for a circuit that was supposedly terminated.
We actually started screenshotting the carrier's own portal showing the circuit status as 'decommissioned' and attaching it to every dispute email. It was the only thing that seemed to stop the automated billing cycle from resurrecting it every quarter.
Data over dogma.