Cutting Cost Per Lead by ~90% With Two Hours of Work

Context

Entyre Care helps family caregivers get paid through state programs. It is a genuinely useful product, and a hard one to sell, because the offer sounds too good to be true. If you tell someone on Facebook that they can get paid to care for their own parent, their first instinct is to assume it is a scam.

Paid social was the primary acquisition channel, running across Facebook, Instagram, and TikTok. Spend was significant, running into the tens of thousands per state per month. An external creative agency produced the ads. Cost per lead was sitting above ~$200 and rising, and it was a recurring topic in leadership conversations.

The team worked in two-week sprints. Anyone could bring a hypothesis, explain how they would build it, and define how it would be measured. I used that structure to test something nobody had questioned.

The hypothesis

I was managing organic social at the time, which meant I was reading the comments on the paid ads every day. The sentiment was not neutral. People were openly calling the ads fake.

That was the signal. The creative was heavily produced and increasingly AI-generated, and for an audience already primed to suspect a scam, polish was working against us. The ads looked like exactly what a skeptical caregiver would expect a scam to look like.

So my hypothesis was not about targeting, bidding, or budget allocation. It was about trust:

If we replace agency-produced, AI-heavy creative with organic, emotionally honest content that acknowledges the caregiver's work before it offers them anything, engagement will rise and cost per lead will fall, because the barrier is credibility, not reach.

The build

I kept every variable constant except the creative and the delivery format.

Creative. I edited a short reel myself using footage I had already captured at previous shoots, plus imagery that felt warm and familiar rather than staged. Older parents, families, quiet everyday moments. The copy led with recognition, not with an offer. Something close to "you are doing more than anyone sees" before arriving at "you do not have to do this alone."

Format. I ran it as a boosted post rather than a formal ad. Boosted posts appear in feed as posts, which reinforced the goal of looking like a real person rather than an advertiser. I chose a reel over static creative on a secondary hypothesis that video would earn more reach.

Targeting. Identical to the existing paid campaigns. This mattered. If the audience had changed, the result would have proven nothing.

Measurement. UTM parameters on the destination link so attribution was clean and comparable against the standard campaigns.

Budget. $200 total across two weeks.

Total production time was roughly two hours, about one to edit and one to publish and configure.

The result

The boosted post substantially outperformed the agency creative on both reach and efficiency. Impressions came in several times higher than a comparable paid ad over the same window, and cost per lead dropped from ~$200 to ~$15.

The comparison that mattered internally was not just the CPL number. It was the input cost. One person, two hours, and $200 outperformed a retained external agency and a five-figure monthly media budget.

Metrics are approximate. I lost access to the original dashboards when my role ended and have reported figures conservatively from memory rather than reconstructing precise numbers.

What I took from it

Production value is not the same as persuasion. For a product whose main obstacle is disbelief, polish actively hurts. The more the creative looked like an ad, the more it confirmed the audience's suspicion.

Comments are free qualitative data. The insight did not come from a dashboard. It came from reading what people wrote under the ads for weeks. I later formalized this instinct into a full sentiment analysis, which became a separate project.

Constrain the test, not the idea. Holding targeting and audience constant meant the result was attributable to the creative alone. A cheap, tightly-scoped experiment made a much stronger internal argument than an expensive, messy one would have.