Hard-Won Lesson

Facebook “Didn’t Work” for Four Years. Then We Built a System.

“We tried Facebook. It doesn’t work for contractors.” Every owner says it, and most of them are right about their experience and wrong about the channel. We said it too — and unlike most, we said it with receipts. Four years of them.

Look at the spend column, not the verdict

Here’s what the failure years have in common, and it isn’t the algorithm. The budgets were $14.8K, $20.8K, $2.5K, and $1.7K. No dedicated owner. No content engine feeding the channel — just ads pointed at an audience with nothing behind them. The two middle years were token spends that couldn’t have proven anything in either direction; a $2,500 year isn’t a test, it’s an alibi. We weren’t running Facebook. We were visiting it.

That’s the uncomfortable diagnosis hiding inside most “it doesn’t work” verdicts: the channel was never actually operated. It was sampled — small budget, side-of-desk ownership, recycled creative — and then sentenced based on the sample.

What 2025 changed — three things at once

In 2025 the channel got what every prior year denied it: a dedicated content hire whose job was feeding it, a full rebrand’s worth of fresh creative running through it, and a real budget — $208K, ten times the biggest failure year. Honest attribution note: those three landed together, so we can’t cleanly separate their contributions. What we can do is show you the ledger.

Still not a cheap channel — a working one

Now the part the turnaround-story version would skip. At 26.6%, Facebook still sits above our 10–15% target band — squarely in the investigate tier of the same kill bands that ended our Detroit TV test. Its demos are expensive: $2,214 each in 2026, more than double the company average. The trajectory is what keeps it funded — 364.7% to 34.0% to 26.6%, falling every period — and it re-earns that funding at every review, same as any other channel. It graduated from the kill tier to the justify-yourself tier. That’s a real promotion. It is not a pardon.

The channel didn’t change in 2025. The operating commitment did. “Facebook doesn’t work” was true every year we didn’t work it.

The test before the verdict

Before you sentence a channel, check whether it ever stood trial. Did it have a real budget — enough to buy actual data, not an alibi? A named owner? A content system feeding it, or just ads pointed at strangers? Twelve consecutive months? If the answer to any of those is no, you don’t have a verdict on the channel. You have a verdict on the visit. And the reverse rule held for us too: the moment the channel did get operated, it still had to live inside the same ratio as everything else — which is the only reason the turnaround is worth believing. The full operating approach is here: Facebook marketing for contractors.

Four years of receipts said it didn’t work. One year of actually working it said otherwise. Both ledgers were telling the truth.