Grow slowly, and only on what sells
A campaign that works at a small size can stop working when you push it. This page is the order we would do things in, the rules we would use to cut and to add spend, and a short daily check. It uses words, not numbers, and every rule says how sure we are.
Seven principles, and how sure we are
Read these in order. The first one matters most.
| Principle | How sure | What we know |
|---|---|---|
| Coverage first Matters most | Proven in our data | Leads from the counties on our coverage site sell far better than leads from elsewhere. It lists the counties (and their ZIPs) where we sell most reliably today. Before you add any spend, aim it at the more densely covered areas. We take traffic from all fifty states, so nothing is blocked. |
| Work the priority groups in order: A first, C last | From our experience coverage site guidance | The coverage site groups counties as A, B and C. Group C changes the most, so it is the least steady place to spend. Start in A and move down only when A is working. We could not show a big sell-through gap between the groups after returns, so do not expect a jump from A alone. |
| Edit locations rarely | From our experience | On Meta, changing targeting restarts the platform's learning. Plan your counties, then leave them alone for a good while. Add a group as a separate step, not as constant tinkering. |
| Pace a first test over about thirty days | From our experience | Expect two to four weeks before you find traction. A bad first week is not a verdict, and a good first week is not a result. |
| Judge ads on our sold-lead signal, not the platform's lead count | Proven in our data one ad | On one ad, the lead count the platform reported was far higher than the real leads our system recorded. Send the platform the sold signal and judge on it. See the tracking page. |
| Results get worse as spend rises, so scale slowly | Proven in our data seen on search | We saw results fall as spend went up on search. Take it as advice for every channel: add spend in small steps and watch what happens before the next step. We have not measured the same curve on other channels. |
| Work one change at a time | Untested | If you change the creative, the county group and the spend together, you will not know which one moved the result. This is common sense, not a result of ours. |
The scaling ladder
Climb one rung at a time. Step back down a rung when a result slips. Nothing here is a promise that you will reach the top.
Check coverage
Pick counties from the coverage site, the more densely covered first. Know which group each county is in.
Small test in group A
One or two creatives, modest spend, about thirty days. Send the sold signal back to the platform.
Read the sold signal
Look at what sold, by creative and by publisher, using sub3. Ignore the platform's own lead count.
Cut what does not sell
Pause creatives and publishers whose leads do not sell after a fair test. Keep the ones that do.
Raise spend in small steps
Only once the sold signal is steady. Wait and watch between steps.
Add group B, then group C
One group at a time, as a planned location edit. C last, because C changes the most.
When to cut, and when to raise spend
Simple rules in plain words. We have not tested these as a set, so treat them as a starting point and write down what you change.
When to cut
- A creative or publisher keeps producing leads that do not sell, after a fair test. Pause it and keep the data. From our experience
- Most of your traffic lands outside the covered counties. Move your targeting toward the more densely covered areas before you spend more. Proven in our data
- The platform reports a healthy lead count but the sold signal is thin. Trust the sold signal. Proven in our data one ad
- A rise in spend made the sold signal worse. Go back to the last level that held. Untested
When to raise spend
- The sold signal has been steady for a good stretch, not just a good day. Raise spend in a small step, then wait. Untested
- You are inside covered counties, group A first. Open group B or C only as a planned edit, one group at a time. From our experience
- Your tracking works: the click ID returns and the sold event reaches you. If it does not, do not scale. From our experience
- Never raise spend because the platform says it is learning well. Raise it because leads sold. Proven in our data one ad
The daily check
Five minutes, once a day. If any line fails, fix it before you touch spend.
- Are the counties you target still on the coverage site? The list can change.
- Does the sold signal still reach you, with your click ID attached? See the tracking page.
- Which creatives and publishers (sub3) are producing sold leads, and which are only producing lead counts?
- Did anyone edit locations, budgets or creatives since yesterday? Write down what changed, so you can tell why results moved.
- Does your pre-lander still work on a real phone, on mobile data, from the ad to the hand-off?
- Are your ads still approved, and still in the right special ad category?
Not sure your setup is ready to scale? Go back to Get set up and ask for a review.