Media Buyer Decision Tree

Find your problem below. Each one has 2–3 actions. Numbers with a blue tag have data proof at the bottom. Partner page: Launch Pack — The First $600.

What are you seeing on your account?
Not a problem, a task: set a daily budget →
Ignore these kill-switch labels when judging ads — they are not quality signals: Duplicate · Call back requested · Booked by Company
You see

A new ad with no leads yet

Judge it by impressions, never by days. (Impressions work the same in cheap and expensive CPM markets.)

Wait< 1,000impressions — no verdict yet
Yellow1,000+may swap for a ready pillar ad
Kill3,000+still zero leads — kill it
  1. Under 1,000 impressions: do nothing. Even good ads look dead this early. proof 1
  2. 1,000+ impressions, zero leads: swap only if a proven pillar ad is already made. Never add extra ads on top.
  3. 3,000+ impressions, zero leads: kill it. In our whole history these almost never recover. proof 2
You see

Spam, fake numbers, landlines

Kill switch label: Spam/Bad Number

The ad or the form is letting the wrong people through. This is a plumbing problem, not bad luck.

  1. Same day: check targeting — geo, exclusion rings, and the qualifier question on the form.
  2. Add OTP phone verification to the lead form. Fake numbers cannot pass it. proof 3
  3. If the junk comes from one ad only (other ads clean, nothing booked from it): kill that ad.
You see

No pickup leads

Kill switch label: No response — but setters rarely select it. A lead with no label and no booking usually means No response.

Real numbers, but no one picks up. Could be the leads, could be our side. Check our side first.

  1. Ask the setters: was the first call inside 30 minutes for each lead? If not, that's the problem — not the ad.
  2. Add OTP to the form — verified numbers pick up more. Same fix as spam, on purpose. proof 3
  3. Fewer than ~10 leads total? Wait. Small numbers lie.
You see

Wrong intent leads

Kill switch labels: Changed Mind · Price Shopper · Job too small

People reply, then back out — real homeowners the ad gave the wrong idea about the price, the offer, or what jobs we do. This is the most dangerous signal on this page. Count kills — never percentages. (1 bad lead out of 3 is "33%" and means nothing. The count is what matters.)

Count1–2kills — normal noise on any account
Yellow3rdkill — check ad price vs real offer
Escalate4thkill in first 20 judged leads — same day
  1. What counts: a judged lead = booked, reached-but-not-set, or killed. Leads nobody has worked yet don't count. Duplicate / Booked by Company / Call back requested never count. On a running account, restart the count on the 1st of each month.
  2. 1–2 kills: normal. Every account gets some. Say nothing, keep counting.
  3. 3rd kill: yellow. Check whether the price in the ad matches the real offer, and note which ads the killed leads came from. Report what you find. Don't touch the ads.
  4. 4th kill before the 20th judged lead: escalate to Sean / Griffin the same day — even if it's only lead 7. A trip means look today. It never means pause or change anything yourself. proof 10
  5. Do NOT swap creatives to fix this. No account has ever been cured of it that way. proof 4
You see

Out of range

Kill switch label: Out of range

Leads from outside the service area — targeting is leaking. Three checks, in order.

  1. Location setting must be "People living in this location" — not "recently in."
  2. Exclusion rings around the service radius must be ON.
  3. The form qualifier must name the metro ("Are you a homeowner in the greater Phoenix area?") — never the state. proof 5
You see

Low booking rate (sub-20%)

Usually no label at all (= treat as No response) · sometimes CANCELED (booked, then fell through — that's a setter/confirmation problem, tell Griffin, not an ad problem).

Below 20% is below every market's natural rate — now you are judging the account, not the ads. Check in this order.

  1. First: cost per booked appointment. Under $100 = the account is winning even if booking % looks low. Do not "fix" a winning account. proof 8
  2. Compare to YOUR market's natural rate (Weather Masters ~35%, Boost ~20–25%, Sam's & State ~50%). Far below your market's natural rate = the gap is fixable plumbing — OTP took Paramedics from 32% back up to 55%. The natural rate is where a healthy account settles; a broken account sits under it. proof 7
  3. New account? Use the first-10 rule: 3+ of the first 10 booked = graduate to normal optimization. 1–2 = plumbing audit before any budget increase. 0 = freeze and full audit. proof 6
  4. Unlabeled unbooked leads = No response. Go to the No pickup card — that's the real problem hiding behind most low booking rates.
You need

To set a daily budget

Work backwards from the appointments the client wants. Use YOUR market's natural booking rate — the rate a healthy account settles at. It barely moves once the account works; a broken account sits below it until repaired. proof 9

Appointments wanted /day
3
÷
Market booking rate
20%
=
Leads needed /day
15
×
Your CPL
$23
=
Daily budget
$300

The example is Boost. Market booking rates: Weather Masters ~35% · Boost ~20–25% · Sam's & State ~50%. If the budget can't support the leads needed, lower the appointment target — don't pretend the math away.

Data proof (fleet: 29 accounts, $164k spend, ~6,000 judged leads)

1. Ads with a lead early are only ~2× more likely to end up good than ads starting at zero — half of good ads have zero-lead days at normal budgets. One day tells you nothing.

2. Zero leads at 1,000 impressions: ~15% recover. At 3,000 impressions: ~0% — none in our history became a good ad. "Good ad" = 5+ leads at $55 or less.

3. OTP took Paramedics from ~32% booking to 51–55%. OTP ads book 51% vs 43% for the rest of that account (138 leads).

4. Every account where "changed mind / too expensive / not interested" passed 24% of leads has died (Go Ecco, Turn Key, NC Quality). Healthy accounts: 3–7%. Creative swaps have never fixed it.

5. Turn Key's form said "Attention Arizona Homeowners" — 56% of leads came from outside the service area, and both early appointments were from bad zips. It cost a $1,350 refund.

6. A healthy account goes 0-for-4 about 18% of the time (normal). 0-for-10 happens ~1–2% of the time (real alarm).

7. Boost booked 20% in February, was repaired (targeting + new creative style), and reached 47–50% by spring. Paramedics went 32% → 55% the same way.

8. Boost right now: booking ~30% but $69–79 per booked appointment — its best unit cost ever, at double the volume. Booking % says "look closer"; cost per booked appointment says "pass or fail."

9. Weather Masters has stayed ~30–40% booking across every ad style we ever ran — a healthy account's rate is a property of the market, not the creative. But a broken account sits BELOW its natural rate until repaired (Paramedics 32% → 55% after OTP, Boost 20% → 47% after the fix). Stable does not mean stuck.

10. The 4th-kill counter was replayed against every launch in the lead sheet (6,188 leads, replayed 2026-08-03). All 7 accounts that later died or churned hit their 4th kill before the 20th judged lead: Go Ecco at lead 6, Turn Key at 7, NC Quality at 8, QPS Charleston at 11, Air Mavericks at 12, Act Right at 18, Onestop at 18 — every one, weeks before anyone called it. No clean healthy launch ever tripped it (State Heating's 4th kill came at lead 76; Garden State and Grace never reached a 3rd). The accounts that tripped and survived — Air Synergy, Paradise — are the ones where the fix happened. The counter fires where the fix should start.