Six categories. The work is the same in all of them.

Most shortlists are the same three or four carriers everybody already knows. The regional provider with fiber in the street outside your building never gets a call because nobody in the room has heard of them, so you end up choosing between the options that were easy to find instead of the ones that were actually good. Diversity has the same problem. Two circuits from two carriers can share a conduit, a landing station, or the same subsea cable, and the paperwork will still say diverse.

I source every provider that can serve the address. All of them, including the ones most people overlook because they aren't obvious. Regional carriers, utility fiber, dark fiber operators, smaller transit providers. That's frequently where the best option turns out to be, either on price or because they're the only one with a path that works. I check the physical route before you sign instead of taking the carrier's word for it. And once the order is in, the install date is mine to manage. If it's going to slip you hear it from me the week it happens.

Vendors answer RFPs optimistically. A roadmap item becomes a current feature, an integration that works in one specific configuration becomes supported, and you find out the difference around month four when you're already committed. The bigger risk is the migration itself. Porting numbers across sites, cutover dates, and a support team that was very responsive during the sale and much less so afterward.

Before you shortlist anyone, I grade every vendor answer against their own public documentation and mark it confirmed, caveat, needs verification, or gap. You get that with the gaps still in it, while you can still act on them. Then I run the project myself. I watch the milestones, chase what's quietly slipping, and escalate when the normal channel stops working. On one phone system replacement the carrier's support team went quiet and milestones kept moving. I went around the account team and got their COO on the phone. It was fixed the next day.

Data center work runs on deadlines you didn't set. A lease ending, a contract you're trying to get out of, a facility closing. Missing that date costs you more than the space or the circuit ever will, because something else renews behind you while you're still sitting in the old place.

I source against the real deadline instead of the comfortable one. When the window is tight that usually means finding the one provider who can genuinely hit it rather than handing you three who can't. Every vendor conversation runs through me, so the date is my problem to chase and not another standing item on your week.

The evaluation eats your calendar. Every provider wants a discovery call, then a demo, then a pitch with three of their people on it. You sit through eight of those to learn that two were ever viable. And the quotes don't compare cleanly, so the cheapest number on the page is regularly not the cheapest deal.

I pre-vet every provider against your actual requirements before you talk to a single one of them. What lands on your desk is a short list of finalists who can meet the requirement rather than ones who say they can. I run the proof of concept and bring you the results. Your hours go into judging what came back instead of chasing schedules and sitting through pitches.

Compute is the easy part to buy. What gates these projects is everything around it. Power and cooling at the facility, lead times on capacity, and the security posture, which becomes a real conversation the moment sensitive data goes anywhere near a model. That conversation is a lot cheaper to have at design time than after you've built.

I source infrastructure built for this specifically. Ultra secure environments, AI-ready capacity, post-quantum encryption. I go to the providers who do this as their actual specialty and bring them to the table, rather than adapting a general purpose facility to a requirement it was never designed for.

Managed services is outsourced accountability, and accountability lives in the SLA and the escalation path. Both get read closely during the sale and then never again. Usually the next time anyone opens the SLA it's two in the morning and the person reading it wasn't in the room when it got signed.

I read the SLA and the escalation path against each other rather than separately, because that's where the gaps show up. Communication structure and named escalation contacts get set before work starts. Then I stay reachable, because the provider's escalation path and your actual escalation path are usually two different things.

The vendors pay me. Worth saying plainly rather than burying it, because it means I get paid when a project finishes and stays running, which is the same thing you want.

If you’ve got something in flight, or a renewal coming that you haven’t looked at yet, that’s a good time to talk.