Where buyers actually get hired

The best client list in this business is public: the Meta Ad Library. Every business in it has already decided ads are worth paying for, and many of them are running accounts with visible leaks — the same two ads unchanged for five months, paid clicks landing on a homepage instead of a product page, a dozen campaigns doing one campaign's job. Those businesses don't need convincing that ads matter. They need someone who can read the account they're wasting money on.

The second channel is agencies. Agencies win accounts faster than they can hire buyers, so overflow and white-label buying is a real entry path: you run accounts under their roof, they keep the client relationship. The pay per account is thinner, but the reps are enormous and the work is steady — a legitimate way to stack managed-spend experience fast.

Third: referrals from the creative side. Editors, designers, and UGC creators see accounts from inside — they know whose ads are stale and whose "marketing person" is the owner at midnight. Make yourself the buyer they recommend: easy to loop in, generous with credit, and quick to send work back their way when a client needs creative.

Qualify hard before you pitch. Professional management makes sense once a business spends roughly $3,000–5,000 a month; below that, your fee eats the budget and nobody wins. The pain you're selling relief from is time: an owner managing an account that spends $5,000+ a month typically burns 15–25 hours a week on it. One honest note on marketplaces: Upwork's median buyer rate skews around $25 an hour because the pool is offshore-heavy. Use marketplaces for first reps if you must — then price for direct clients and aim there.

The free audit pitch

Your pitch is a free audit: a one-page teardown of the prospect's account — structure, tracking, creative — naming specific leaks in plain language. It works because it demonstrates the exact skill you're selling, on their actual numbers, before they've paid you anything. What it never contains is a promised result. You can say "this structure is wasting spend." You cannot say "I will get you a 4.0 ROAS." That line, from Module 0, is the honesty standard of the whole job.

Example DM (to the owner of a DTC skincare brand): "Hi Lena — I found Glowbar through the Meta Ad Library. Your vitamin C serum ad has been running unchanged since March, and it lands on your homepage instead of the serum page, so the click you paid for has to go hunting for the product. I wrote up a one-page teardown of the three leaks I'd fix first, including the campaign structure — eleven campaigns where three would do the same job. Want me to send it? If it's useful, I buy ads for brands like yours on a flat monthly fee, with a percentage once spend grows. Either way, pointing that ad at the product page is free money. — [Your name]"

Example email: Subject: "The page your serum ad is paying to send people to." Body: "Hi Lena — Glowbar's serum ad has run unchanged since March, and it sends paid clicks to your homepage. Three things I'd check in week one. First, tracking: the pixel alone misses conversions, and without the Conversions API backing it up you're reading a partial picture and optimizing toward it. Second, structure: eleven campaigns where a consolidated three would give the algorithm cleaner signals to learn from. Third, retargeting: there's no exclusion for recent buyers, so people who purchased last week are eating prospecting budget. I buy ads for DTC brands on a flat monthly fee — $1,200 at your current spend — starting with a 30-day pilot and one success metric agreed in writing. Worth a 20-minute call Thursday to walk through the teardown? — [Your name]"

Send it to a named person, keep it to one page, and follow up once — a week later, with one new observation, not a nudge. Then stop. An audit this specific either starts a conversation or it doesn't; volume was never the point.

Trial structures that protect both sides

A pilot is a defined product, not a free sample. The standard shape: 30 days, a ring-fenced test budget the client caps in advance, a flat pilot fee (one month at your entry-to-mid retainer band, $500–2,500 depending on the account's size), and one success metric agreed in writing before launch — for example, "CPA at or under the $48 break-even on the hero product, at $3,000 of test spend." One metric, not five. At the end of 30 days you both know exactly what "it worked" means, because you wrote it down when nobody had anything to prove.

See what this protects. The client risks one month and a capped budget instead of signing a long contract with a stranger. You get paid for real work and produce a clean before-and-after instead of a vibe. What a pilot is not: unpaid management. The audit is free because it's an hour of reading. A month of buying is the job itself, and doing it free teaches the client your work is worth nothing. It's also not open-ended — the day-30 review meeting goes on the calendar on day one.

Put the pilot terms on one page: the test budget and who can approve changes to it, the metric and where it's measured, what the fee covers, and what happens at the end. The client owns the ad account and its data — always, pilot or not. If it ends, the campaigns stay, documented, and you hand over a short summary of what was tested and learned. Buyers who exit clean get referred. Buyers who hold accounts hostage get talked about, accurately, forever.

Onboarding: access, tracking, expectations

Start with access, done properly. For Meta, the client adds your business as a partner in their Business Manager; for Google Ads, you link your manager account to theirs. Never ask for their personal password. Shared logins trip security checks at the worst moments, stop working the day the client changes a password, and blur who's responsible if money moves oddly. Partner access is also how the relationship ends cleanly — they revoke a permission instead of changing every password they own.

Next, the tracking check, before you change a dollar of spend. Fire test events and confirm the pixel records the purchase. Confirm the Conversions API is sending the same events from the server side. Then agree, in writing, on the one conversion definition you'll report against and the attribution window that counts it. Most accounts that "don't work" are measurement problems wearing a media costume: the ads were fine, the scoreboard was broken, and every decision made from it was fiction.

Then the expectations document — one page, plain words. What you control: account structure, budgets, testing cadence, pause decisions. What you don't: their offer, their prices, their stock levels, their website's speed. Plus the reporting date each month, how fast you'll reply on business days, and how budget increases get approved, so the first surprise a client ever gets from you is never a spend jump they didn't sign off on.

Shape the first two weeks conservatively: fix tracking, consolidate the structure, and keep the current winners running while your first tests spin up beside them. A dramatic day-one overhaul feels decisive and destroys the baseline you need to prove anything later. Month one is for a trustworthy scoreboard and early, explainable wins.