The exit test
Ask one question before you sign anything: what's still mine the day after I stop paying? That answer grades a marketing engagement better than any case study, any pitch deck, and any list of deliverables.
It works because it's hard to fake. A vendor can describe the same work either way, but ownership is a yes or no per asset, and the answers are checkable in an afternoon. Run the test before you sign, while the terms are still open. Nobody has ever improved their ownership terms during a cancellation.
The asset inventory
Six things get built during a serious marketing engagement. Go down the list and mark each one yours or theirs.
- The website: the design, the templates, the code, and the hosting account it runs on.
- The content: pages, posts, landing pages, and the briefs and outlines behind them.
- The frameworks: your keyword and entity maps, the topic architecture, the testing backlog, the prompt sets behind any AI measurement.
- The accounts: analytics, search consoles, ad platforms, tag manager, business profiles, the domain itself.
- The data: historical performance, conversion and revenue records, and the joins that connect them.
- The dashboards: the reporting layer where all of the above gets read.
How a retainer usually keeps them
Rarely by writing 'we own this' into a contract. It happens through defaults nobody argued about at signing.
The site gets built on the agency's hosting, under the agency's license, on a page builder that only renders inside their stack. Analytics and ad accounts get created by the agency, on agency logins, with you added as a user instead of an owner. Reporting lives inside their platform, so your history is a login you lose. The keyword map and the content calendar exist as working documents in their drive, and the version you see is a monthly PDF.
None of that is necessarily malicious. It's just what happens when a vendor optimizes for running many accounts efficiently. But the effect is the same: leaving costs you the work, so staying stops being a decision you re-make.
Account ownership from day one
Accounts are the cheapest thing to get right and the most expensive thing to discover late, because account history is the one asset that cannot be rebuilt. Two years of search data has no replacement. You either own it or you start over.
Check every one of these has your company as the owner, and an internal person as an admin, before work starts:
- Analytics: your property, your organization, agency users added as editors.
- Search consoles, Google and Bing both: verified to a domain you control, not a vendor tag.
- Ad accounts: your billing, your account, agency access granted by permission.
- Tag manager: your container, your published history.
- Business profiles and listings: your logins, and the recovery email on a company address.
- Domain and DNS: registered to your company. This one gets missed more than any other.
- CMS and hosting: an owner-level account for someone on your payroll.
Why month-to-month only works when you own the assets
A no-contract engagement sounds like a favor to the client. It only functions when the ownership question is already settled, and that is the part most people miss.
Think about what makes a client stay when there's nothing stopping them from leaving. Either the work is producing results you can see, or leaving is too painful to consider. Those are the only two retention mechanisms. A contract sells the second one. Ownership plus visible results is the first.
That's also why the two go together and can't be split. Month-to-month without ownership is worse than a contract, not better: you can leave any month, and every month you leave with nothing. A twelve-month contract where you own everything is at least honest about the trade.
The compounding argument
Owned infrastructure gets more valuable every month. Rented deliverables reset to zero the day you stop paying, and the reset is total, not partial.
A page you own keeps earning. A keyword framework you own gets refined instead of rebuilt. Two years of clean account history makes the third year of analysis better than the first. An entity map you own carries over to the next agency, the next platform, and the next channel that hasn't been invented yet.
Compare that to the alternative. Three years of a rented engagement produces three years of reports, and then a fourth year that starts from a blank analytics property and a site you have to replace. The spend was real. The asset is not there.
This is the whole argument for treating marketing as infrastructure rather than as a service you buy monthly. Not because infrastructure sounds more serious, but because assets appreciate and deliverables expire.
Five questions to ask before you sign
Ask these in writing, and read the answers for specifics rather than reassurance. Anything vague is a no.
- Which accounts will my company own outright, and who at my company will hold admin? Ask for the list, not the principle.
- If we part ways, what exactly transfers, in what format, and how long does it take? A good answer names files and timelines. A bad one says everything.
- Where does the site live, on whose license, and can my developer maintain it without you? If the site only renders inside their platform, you are renting the site.
- Where does my reporting history live, and can I keep it? If the dashboards are their product, your history is their product too.
- What did the last three clients who left take with them? Anyone who has actually built it this way can answer immediately.
What this looks like when it goes right
One example from our own work. A law firm came to us needing a full site migration, having lost rankings on a previous redesign that took two years to recover. The hard requirement wasn't growth. It was lose nothing.
Nothing that ranked before cutover dropped in the ninety days after it. The part that matters for this piece is what the firm holds now: the site, the redirect map, the practice-area content, the tracking accounts, and the rank history through the migration window. If they replaced us tomorrow, the next team would start from a documented system instead of an archaeology project.
That is the only version of this argument that means anything. Ownership is not a clause. It is a list you can check.
Case metrics are illustrative placeholders pending client approval to publish named results.