Six mistakes account for almost every failed first year of small business advertising in the US, and exactly one of them is about which tool you bought. They are listed here in the order they cost the most, which is deliberately not the order people worry about them in.
Mistake 1: not knowing what a customer is worth
The most expensive mistake and the least discussed, because it is not a marketing problem — it is an arithmetic problem nobody made you do.
Without this number, no result can be judged. A $200 cost per lead is outstanding if a client is worth $15,000 and ruinous if an average ticket is $80. People without the figure default to judging on cost per click, which is the metric most likely to mislead them.
Work out two numbers
What a customer is worth to you over their lifetime, and roughly what share of enquiries become customers. Fifteen minutes with your own records.
Then the one that matters
Cost per lead divided by close rate equals cost per customer. That is the figure to hold against lifetime value — everything else is a proxy.
Why no tool supplies it
It lives in your books and your head. This is the input that makes every automated decision downstream interpretable, and it is the one part nothing can automate.
Mistake 2: the wrong channel for how your customers find you
Most US small businesses start on Meta because the creative is satisfying and the interface is familiar. For roughly half of them that is the wrong first channel.
They search for you
HVAC, plumbing, legal, dental emergencies, locksmiths, auto repair. Google first, without question. Nobody scrolls Instagram and decides their furnace should fail.
They discover you
Restaurants, med spas, boutiques, fitness studios, events. Meta first. Nobody searches for a nail salon they have never heard of.
Genuinely both
Home improvement, weddings, anything researched for weeks. Start with search, add social once search is producing.
Getting this wrong is not a small inefficiency. Running social for an emergency service means paying to interrupt people who do not currently have the problem, which is close to unwinnable regardless of how good the ad is.
Mistake 3: giving away the ad account
This is the one that is invisible for two years and then costs you everything at once.
If your campaigns run in an ad account owned by an agency or a software vendor, then your pixel history, custom audiences, conversion data and spend record belong to them. Leaving means starting from zero — and in the US that history is expensive to rebuild and materially improves platform performance while you have it.
💡 Ask one question before connecting anything to anyone: do campaigns run in an ad account I own, with you granted access? Get the answer in writing. It is trivially easy to agree to at the start and very hard to unwind later, and it is the difference between being a customer and being stuck.
Mistake 4: buying the cheapest attention available
US auctions are expensive because they work — the price reflects what the audience is worth. A cost per click far below what your category commands usually means you found people who are cheap to reach for a reason.
Two diagnostics worth more than cost per click:
- 1The gap between clicks and landing page views. On one campaign we watched 45 clicks produce 28 arrivals — nearly 40% tapped and left before the page rendered. That is traffic quality, not site speed.
- 2Whether anything downstream moved. Form starts, calls, add-to-carts. If nothing after the visit happens, the click price is irrelevant and the audience is wrong.
Cheap attention that never converts is the most expensive thing on an account, because it also teaches the platform the wrong lesson about who to show you to — and that lesson outlives the campaign you learned it on.
Mistake 5: measuring the conversion that is easiest to measure
Most US service enquiries are phone calls. If your tracking counts only form submissions, you are measuring the minority of your leads and instructing the platform to optimise toward it.
Set up call conversion tracking with a minimum call duration so wrong numbers do not train the algorithm, and use call extensions so the number is tappable straight from the search result — a large share of callers never reach your website at all.
And verify the event fires at all. A campaign optimising toward a conversion that never happens is optimising toward nothing, and produces no warning of any kind.
Mistake 6: reacting in week one
The most common, and it looks exactly like diligence.
Hours 0–5
A new click or conversion ad set takes four to five hours before its first impression. Reach objectives start in minutes, which is why comparing them makes yours look dead.
Days 1–3
Data exists, means nothing. The platform is learning who to show it to. Early cost per result is not your cost per result.
The cost of intervening
Every edit to a live ad set restarts the platform's evaluation. Change things four times in a morning and the ad set never gets far enough in to deliver properly.
The only week-one interventions that are justified: a rejected ad, a billing problem, or a conversion event you discover is not firing. Everything else waits a fortnight.
And the one that is about the tool
Whether it does the mechanical work correctly — objective matched to your goal, audiences built as comparable segments, keywords from real search volumes rather than generated, budget above the platform floor, and everything shown to you before it publishes.
That last point is not a preference in the US. AI writes the exact claims the FTC treats most seriously — outcome guarantees, invented statistics, testimonial-style copy from people who do not exist — and the advertiser stays responsible. A tool that publishes without showing you first has removed your only review step.
Your ad account, your data, previewed before it runs
Adyft builds Google and Meta campaigns from a short description of your business and launches into accounts you own. 14-day free trial, no card, 0% of ad spend.
Start free — no card needed