Plumbing Company
Leads come from everywhere — search, vans, referrals, a directory, the radio — and the owner can’t say which actually book jobs. We model the mix and move the budget to what pays.
The situation
A plumbing company runs a dozen marketing channels and judges them by gut and by whichever salesperson shouts loudest. Some channels look busy but book low-margin drain calls; others quietly drive the high-value repipes and water-heater installs. Without a model, budget flows to noise and the profitable channels go under-fed.
Attribution is muddied because customers see several touchpoints before they call, and emergency demand is partly driven by weather and seasonality that has nothing to do with marketing — so raw lead counts mislead.
Where we dig for the truth
We build a marketing-mix model that separates each channel’s true contribution to booked, profitable jobs from seasonality and overlap.
Local search and the branded vans quietly drive most profitable jobs; the directory and radio absorbing real budget barely move revenue once overlap is removed.
Our approach — Marketing-Mix Modeling (MMM)
Marketing-mix modeling uses weeks of spend and results to estimate each channel’s incremental effect on booked, profitable jobs — controlling for weather, seasonality and the fact that channels overlap. Each channel gets a true return, and diminishing-returns curves show where the next dollar stops paying.
Budget shifts from low-return channels to the ones that drive high-margin work, and spend is re-weighted toward the job types — installs, repipes — that actually carry the business.
Cutting the directory and re-weighting toward search, vans and referrals roughly halves the blended cost of a booked, profitable job.
What changes
Same total budget, allocated by model. Representative for a mid-size plumbing company.
Frequently asked questions
How do you know which marketing channels book plumbing jobs?
What is marketing-mix modeling (MMM)?
Isn't more advertising the answer?
Want this run on your numbers?
Share your channel spend and booked-job data and we’ll model what actually pays.