Fewer leads on purpose.

Situation.

Whole-home insulation is a high-ticket local sale. The client could spend up to $2,000 to land a customer and was aiming for a 6x return, so a cheap lead that never closes does nothing for that math.

What I did.

1. Lead gen in 18 local markets, with a form that qualifies.

   I ran Meta lead gen across 18 local markets and made the form do some of the qualifying. Prequalification and two to three screening questions added friction on purpose. In our best market, cost per lead ran $25 to $35.

2. Creative that matched the season and the market.

   Creative spoke to the season and the market: Detroit winter, summer heat and humidity, and the tax credit offer. Copy carried a location placeholder, so each market's ad named its own city or area.

3. Offers with a reason to book.

   The copy docs show the offers the ads were built around. For winter 2023 it was no interest for 72 months on whole-home packages. For Black Friday and Cyber Monday 2023 it was free labor, a free Nest smart thermostat, and a $1,200 instant cash back tax credit, for appointments booked in November. The education angle was that 77 million homes have inferior insulation.

4. A testing log for every market.

   Every market had its own tab in the testing log. Each test got an ID, its dates, how it was set up and bid (ABO or CBO, cost cap, cost-per-result goal, or highest volume), a result tag, and a one-line conclusion. The tags ran from Winner and Hopeful to Loser, Not enough spend, and Awaiting result.

The creative.

Statics and one reel from the account. The portfolio has a line on what each one was for.

More context on each piece is in the portfolio.

What the log showed.

Tests ran from February to May 2023. These are the patterns that held up.

• Localized copy gets the spend. Testimonials earn it back.

  When localized copy and testimonial copy ran against each other, the algorithm gave localized copy far more of the spend. Testimonial copy did well whenever it got spend. The call in the log: blend localized copy into almost everything.

  Log summary and Knoxville test 1, Feb 15 to 23, 2023
• A foam-fill reel drove the lead volume.

  In Detroit, a foam-fill reel with a broad audience on a cost cap generated 74 leads between March 23 and April 19, 2023, and was logged a clear winner. The Charlotte notes say the foam-fill reel delivered most of the lead volume there.

  Detroit test 5 and Charlotte notes
• Seasons expire copy and audiences.

  By spring, Detroit's winning interest stack stopped working, most likely because of cold-weather interests in it. In May a test was logged "copy too cold weather related." The next round used summer creative with scarcity copy and was tagged a winner.

  Detroit tests 7 to 10, spring 2023
• Bid controls need watching.

  A cost-per-result goal won a bid test in Tri-State and could not get enough delivery in Grand Rapids. The Lansing note: a cost-per-result goal can be very effective, but watch delivery and pacing.

  Tri-State, Grand Rapids, and Lansing, Feb to Mar 2023
• Cutting budget killed winners.

  In Mid Michigan, ad sets that were working stopped delivering once spend was reduced. The log's note: ABO worked really well, but reduced spend killed the account.

  Mid Michigan, Mar 27, 2023

Takeaway.

The target was a customer at $2,000 or less, so I built the form around that number instead of chasing the cheapest lead.