Work / Paid social

Builder Funnel · 2026

Rebuilding paid social around booked calls.

Meta was bringing in the cheapest leads in our account and almost none of the customers. Our booking campaign had drifted up to about $3,200 per booked call, so I rebuilt the account around the one thing sales cares about: calls on the calendar.

Meta adsFunnel mathHubSpotLead forms
$3.2K$493per booked discovery call across October, down from about $3,200 on the old campaign by July
0 → 4booked calls from the form-first campaign, week one ($579) to week two ($853)
18discovery calls booked from Meta in October, on $8,875 in spend

TL;DR

  • From January to August, Meta brought in 340 leads and 2 customers. The ads were optimizing for cheap leads, so that's what we got.
  • I killed a direct-to-booking campaign that spent about $1,027 with 0 real bookings, and replaced it with a short form first, then the calendar.
  • Across October, Meta booked 18 discovery calls on $8,875, or $493 per booked call.

The problem

Our old cold campaign sent people straight to a booking page. For a while it worked: calls cost about $819 each through late March. Then it slid, and by July a booked call cost about $3,200. Its blended cost for the whole run made it look like it had never worked, which hid both the good months and the moment it turned.

Underneath that, Meta's leads were the cheapest in the account and turned into the fewest customers. We were telling Meta to find people who fill out forms, and it did exactly that.

What I did

  1. Rebuilt the account. Four campaigns at $316 a day, each with one job: a Growth Math Calculator lead magnet I built as a HubSpot module, a quiz, a proof campaign and a direct-to-booking test.
  2. Cut the quiz down. From 47 ads to 8 and from about $298 a day to $80, then added a revenue qualifier.
  3. Killed the booking test. About $1,027 in, it had 0 real bookings. At roughly one booking a week, Meta has nothing to learn from.
  4. Launched form-first. A three-step form, then the rep's calendar opens already filled in. I moved the Lead event to step one so Meta had a signal to optimize on.
  5. Cleaned up placements. Audience Network was taking about 56% of spend and sending junk. I excluded it everywhere.
Step 1Name, email, phone, company↑ Meta's Lead event fires here
Step 2Location and company revenue
Step 3Monthly marketing spendplus "What do you want out of the call?" (optional)
ThenThe rep's calendar opens, already filled in
The form-first flow. Asking for a booking up front gave Meta about one signal a week. A short form first gives it one every time someone starts.

What happened

The form-first campaign booked 0 calls on $579 in its first week, then 4 on $853 in its second. Across October, Meta booked 18 discovery calls on $8,875.

Cost per booked discovery call, by week

Oct 5-11 (4 booked)
$562
Oct 12-18 (2 booked)
$1,099
Oct 19-25 (3 booked)
$714
Oct 26-Nov 1 (9 booked)
$255
Meta spend across the calculator, form-first and quiz campaigns, divided by discovery calls booked that week. From Builder Funnel's weekly ads reports.
Watch the weekly number. A running average blurs the turn. Cost per booked call week over week shows it the week it happens, so that's the number I watch.

The quiz got its own fix. A revenue question raised the share of $1M+ companies among quiz leads to 50% in October. In November I paused it and moved its $40 a day to form-first.

What I'd do differently

I'd build the proof campaign first. The plan called it the piece that makes the other campaigns pay, and it's the one that never got built.

I'd also keep test bookings off the live pixel. My own test booking taught the booking campaign the wrong people right when it needed real ones.

Bookings went up, and now show rate is the next thing to fix.

Built withMeta Ads ManagerHubSpot forms + workflowsClaude CodePython