The electrical licence is six years younger than the air conditioning one. The ad account was treating them as the same business

Who this is

Reeis has been working in Phoenix since 2009, selling air conditioning, plumbing, electrical and insulation. Three registrations with the Arizona ROC: B-318364, CR39-318363 and CR-11-354282.

Read those numbers in order and the company's history falls out of them. The first two were issued together. The electrical one is roughly thirty-six thousand later in the sequence, which is several years in a register that issues them in order. Air conditioning is the founding trade. Electrical arrived long after the habits had set.

What a trade inherits when it arrives late

Nobody sat down and decided electrical should be advertised like air conditioning. It simply got added to an account that already worked, and an account that already works has opinions baked into it: what a good cost per lead is, what a landing page looks like, which hours are worth buying, how long to wait before calling a keyword a failure.

Every one of those opinions was calibrated on a trade with a different ticket. In Phoenix an air conditioning replacement and an electrical service call are not the same transaction, are not found with the same words, and do not survive the same bid.

An electrical job that costs less to win than an air conditioning job is not automatically a better job. It is a smaller one. The mistake is judging both against a single target.

The vocabulary problem

Electrical search in Phoenix splits into work that is urgent and cheap, work that is planned and expensive, and a large amount of traffic that is neither because it belongs to somebody with a multimeter.

The account was buying all three at one bid. The planned work — panels, service upgrades, car chargers — is the part that pays, and it was being outbid by the urgency of a tripped breaker at two in the afternoon.

What we were unsure about

Splitting electrical out means it can no longer borrow budget from air conditioning on a slow day, and a smaller trade with its own budget can stall. That risk is real and we did not have a way to size it beforehand.

The thing that made it safe enough to try was that the ticket on planned electrical work is high enough to survive a slower learning period. If the trade had been all forty-dollar service calls we would have left it where it was.

What came back

Fewer electrical enquiries, and almost twice as many electrical jobs. The top of the funnel narrowed deliberately when the diagnostic searches came out.

The row that matters is the bottom one. Everything above it is bookkeeping.

The part that had nothing to do with advertising

A quarter of the improvement is a landing page that prints a price range for a panel upgrade. Not a quote — a range, with the reasons it moves.

Contractors resist this, reasonably, because a range invites an argument. What it also does is end the calls from people who were never going to spend that much, and those calls cost a technician's afternoon whether or not anything is sold.

What this does not prove

• Ninety days is a short window for planned electrical work, where the gap between the first search and the signed job is often longer than the window itself. Some of the jobs counted here were started before it.
• Electrical is the smallest of the three trades in this account, so the percentages move on smaller numbers than the air conditioning figures would.
• The landing page and the campaign split went live in the same fortnight. We report them together because we cannot separate them honestly.
• Cutting the diagnostic searches removes future customers as well as non-customers. We accepted that trade at this budget and it would be the wrong trade at a much larger one.
• Insulation is also part of this business and is not covered here at all. It has too few jobs in the window to say anything about.

What to take from this

1. Judge each trade against its own ticketA single target cost per lead across a multi-trade business quietly defunds whichever trade has the smaller average job, regardless of how profitable it is.
2. Separate planned work from emergency workEmergencies convert faster, so in a shared campaign they take the budget every time, and the planned work is where the margin lives.
3. Publish a range for your highest-ticket jobThe calls you lose are the ones that were never going to buy, and every one of those costs you a technician's afternoon to discover.
4. Read your own licence numbersThey are issued in order, so they tell you which trade is oldest — and the oldest trade is the one whose assumptions all the others quietly inherited.

Questions people ask about this one

Why does a newer trade end up mismanaged inside a working account?

Because a working account is full of settled decisions — target cost per lead, bid strategy, page structure, which hours to buy — and all of them were tuned on the founding trade. A trade added later inherits every one of those decisions without anybody re-examining them. The account is not broken. It is correctly tuned for a different business.

Should electrical have a lower cost per lead than air conditioning?

Usually yes, and that is exactly why a single target is dangerous. An electrical service call is a smaller ticket than a system replacement, so it has to be won for less to be worth the same. Holding both to one number means either the electrical work is starved or the air conditioning work is overpaid for. Each trade needs a target derived from its own average job.

Does publishing prices lose you customers?

It loses you calls, which is not the same thing. In this account the enquiry count went down and the sold job count went up, because the people who stopped calling were the ones who would have ended the visit at the number. A range with the reasons it moves is also a fair thing to give someone before they take an afternoon off work.

How long before a separated trade starts performing?

In this account the electrical campaigns were noisy for about five weeks before the cost per sold job settled. That is the real cost of separation and it should be budgeted for. A trade whose ticket cannot support five slow weeks is probably better left inside the main account until it can.