Two businesses I know well almost lost everything this year. Not to a market crash or a bad quarter. They lost it switching agencies, because the people who built their marketing stack held the keys, and they never did.
I’m talking about the whole machine: the website, the ad accounts, the automations, the pixel data, the AI signals, years of performance history. Most of it gone, or one expired password away from gone. The kind of thing you assume is in your name right up until the day it very much is not.
I get why it happens. You hire experts so you don’t have to think about hosting logins and pixel setups. That’s the point of hiring them. But “I don’t want to think about it” and “I don’t own it” are two different things, and the gap between them is where the real risk lives.
Let me walk you through what I actually saw, then give you the audit I now run on every stack I touch.
The website that went dark, and the math nobody ran
A company signs with a new agency, ready to grow. Then the website goes down. No warning, no access, no backup.
Here’s the chain of events. The old agency owned the hosting account. They let the subscription lapse. They never transferred ownership. So when the renewal failed, the site went with it, and the new team had no login to save it. In a scramble they pushed an old version of the site back up just to stay visible. The real, optimized site stayed stuck.
Now the part that stings. Run the math on two weeks dark:
- Roughly 1,000 visitors lost
- 16 qualified leads that never arrived
- A quarter of those leads usually close
Call it $40,000 in deals gone over fourteen days. Around $2,800 a day, every day the site sat dead. Your numbers will differ, but the shape holds: downtime you can’t fix because you don’t have the keys is pure, compounding loss.
This was never a tech problem. It was a business risk hiding inside an org chart, waiting on one question nobody asked: do we actually own this?
The funnel held hostage by one Gmail
Different company, same root cause, sharper edge. They’d hired a freelance contractor to build out the tech stack, and the work was genuinely good:
- Automations wired up in Zapier
- Pixel and conversion tracking across Meta and Google
- CRM integrations and data flows that actually flowed
Every piece of it lived under the contractor’s personal Gmail. When the project wrapped, they were one password reset away from losing the entire funnel, and they had no idea how exposed they were until someone outside the project pointed it out. The build was great. The ownership was a landmine.
These aren’t tech slips. They’re business risks.
This is the part that gets me. You’re doing everything right. Growing the team, investing in digital, hiring real specialists. But if you don’t own the systems those specialists build, you’re scaling on someone else’s foundation, and you’ll find out exactly how solid it is at the worst possible moment.
It bites harder in the AI era, because so much of your performance now improves off your own data signals. Lose the infrastructure and you don’t just lose access, you lose the learning. I’ve watched versions of all of these play out:
- Six figures of ad testing that vanished with an account
- AI recommendations that stopped getting smarter because the data feeding them walked out the door
- Teams burning weeks re-learning what the pixel already knew
- Whole systems rebuilt from scratch because nobody documented the first build
Ownership of your stack is an architecture decision, the same “what do we own versus what do we rent” call I walk through in your AI marketing stack. Renting the work is fine. Renting the keys is not.
The ownership audit: are you actually in control?
Here’s the checklist I run. Answer no to any of these and you’ve found something to fix this week:
- Do you own your domain and your website hosting, in your account?
- Are your ad accounts (Google, Meta) registered under your business, not an agency’s?
- Do you have admin access to GA4, Google Tag Manager, and Search Console?
- Are your CRM and email tools (HubSpot, Klaviyo, whatever you run) under a company account?
- Are Zapier, Make, and your AI tools tied to company emails, never a freelancer’s personal one?
- Is your documentation saved somewhere central and company-owned?
- If your best contractor left tomorrow, would your systems keep running without them?
That last one is the whole test, really. If the answer is no, the rest of the list is just detail.
When you offboard an agency or freelancer
The handoff is where ownership gets won or lost. Before anyone says goodbye:
- Change every shared password
- Transfer ad account ownership, not just access (those are different, and the difference matters)
- Export all campaign data and your custom audiences
- Back up CRM flows, pixel events, and your tag manager setup
- Get written SOPs for any custom automation or AI logic
- Document what was built while the person who built it is still reachable
You will not remember how it all fits together in six months. Nobody does. Write it down now.
Treat your stack like a financial account
If your marketing systems are humming but you don’t own them, you’re one breakup away from a mess you didn’t see coming. Your data is leverage. Your tech stack is equity. Treat both the way you’d treat money in the bank: secure, transparent, and in your name.
None of this means you stop hiring help. I’m all for bringing in sharp people who build things you can’t. It means you hire them onto a foundation you own, so the value they create stays yours when they move on. That’s the difference between a stack that compounds and a hostage situation with a calendar invite.
If you want the bigger picture on which parts of your marketing to own outright, which to rent, and who’s accountable when a vendor changes the deal, start with where AI actually fits across your marketing and then go deep on your AI marketing stack.
Get the ownership audit, plus the next quiet risk, in your inbox
Every week I take apart one piece of marketing infrastructure: what to own, what to rent, where the quiet risks hide, and the exact checks I run before I trust a stack. If this post saved you from one ugly handoff, the next issue will flag a few more before they bite.