Somewhere in your business there is a login you don't have. A Google Ads account a former employee built and nobody turned off. A Meta page a relative set up years ago. A website a freelancer handed over with a folder of instructions no one opened. Something in that pile is spending money this month. Something else in it is quietly working. You cannot say which is which, and that is the actual problem — not a blank page, but a half-built machine with no manual and no owner.
Which is why the useful first move isn't picking a vendor or a channel. It's taking stock. What you find decides whether you need a pair of hands, a specialist in one channel, or somebody to make decisions — three different jobs that get sold under the same words and cost wildly different amounts.
Make the list nobody has made
Open a spreadsheet with four columns: the channel or tool, who holds the login, what it costs per month, and the date someone last looked at it. Then fill it in — every ad account, every subscription, the domain registrar, the email tool, the booking widget, the review platform, the agency retainer that renews silently. This takes an afternoon, and it routinely turns up two or three line items nobody can justify. Cancel those before you spend a dollar on anything new; you have just funded part of whatever comes next.
Match the symptom to the fault
| What you're seeing | The likely fault | The kind of work that fixes it |
|---|---|---|
| Ads run, the phone doesn't ring | Wrong search terms, or the wrong page after the click | Campaign rebuild and landing page work |
| The phone rings, but you can't say from where | Nothing is measured end to end | Conversion tracking and analytics setup |
| Leads arrive and none of them fit | You're paying for curiosity, not intent | Keyword, audience and negative-list cleanup |
| Costs climbed, volume didn't | Auction pressure, or budget spread across too many campaigns | An account audit before any new spend |
| Nothing is running at all | No clear offer, or no capacity to serve one | Positioning and pricing — not media buying |
The price of the click is a warning label
This is a small, expensive market to advertise in. Roughly 460 searches a month across the whole cluster, made of 8 distinct queries that mean slightly different things, and advertisers pay as much as $64.57 for a single top-of-page click on it. Demand also swings hard — 480 searches in September 2025 down to 110 in July 2026. Read that as a caution about the results page you're standing on: when one click costs that much, only firms with a serious sales operation can afford to be there, and polish in the pitch tells you about their marketing budget, not their competence with yours.
Count what you already have before you buy more
Most owners in this position are one broken measurement away from being able to answer their own questions. If a form fill, a phone call and a booking each land somewhere you can see — with a source attached — then the inventory you just built stops being a list of costs and becomes a list of results. That reordering matters: fixing tracking is cheap and permanent, while buying traffic you can't attribute is expensive and repeats monthly. If you do only one paid piece of work this quarter, make it this one.
Three kinds of help, three different bills
A freelancer or contractor is hands. They execute what you specify, cost the least, and are the right call when your inventory shows one clear broken thing and you know what "fixed" looks like. A specialist agency goes deep on one channel and brings its own measurement; that's the fit when one channel carries most of your revenue and it's underperforming. A fractional marketing lead makes decisions across everything — the expensive option, and the correct one only when your list has several live channels, no owner, and a history of things being started and abandoned. If your inventory came back short and messy, hiring a lead is overpaying for a problem you can still solve yourself.
The next two weeks
Week one: finish the spreadsheet, recover the logins, kill whatever nobody can explain. Week two: submit your own form, call your own number, and follow one lead all the way to where you'd record it — then check whether anything about that journey shows up in your analytics. Whatever fails in week two is the brief. Hand that brief, plus the spreadsheet, to two vendors and see who asks better questions about it. You may find you don't need either of them yet; you'll certainly stop paying for the version of this conversation where nobody knows what's already running.