What proving ROI requires
Four numbers, joined up:
- What you spent, by channel
- What enquiries arrived, attributed to those channels
- Which of them became jobs
- What those jobs were worth
Most small businesses have the first reliably, the second partially, and the third and fourth not at all in any form that can be joined to the others.
The third is the one that breaks it, and it is the one only you can supply.
Why the third number is missing
It lives in people's heads, in a scheduling app, or on invoices with no source recorded. So the moment an enquiry becomes a job, the link back to what produced it is severed.
Fixing it is not technical. It is adding one field — where did this come from — to whatever record you already keep, and filling it in.
What can be done retrospectively
More than you might expect, if the raw material exists:
- Match invoice dates against enquiry dates
- Use tracking number data to identify which calls came from where
- Ask customers, for recent jobs — most will tell you
- Reconcile ad spend periods against enquiry volume
This is manual, imperfect and still worth doing. A rough figure that is roughly right beats a precise figure about the wrong thing.
The comparison that matters
Not return on ad spend in isolation. Cost per booked job by channel, against your value per lead.
A channel producing jobs at $180 each when a job returns $400 in gross profit is working. A channel producing jobs at $600 is not, no matter how good its click-through rate looks.
The lifetime value adjustment
If customers come back — maintenance plans, annual service, repeat repairs — first-job value understates what a channel produced.
A channel that looks marginal on first-job value can be clearly profitable on lifetime value. Businesses that ignore this systematically under-invest in acquisition and cannot work out why competitors outspend them.
What honest analysis will not do
Produce a clean single number. Attribution across channels is genuinely difficult — a customer who saw an ad, then searched, then asked a neighbour, is not cleanly assignable.
Anyone offering exact attribution is overselling. What good analysis produces is a defensible ranking of channels and a clear read on which are obviously working and which obviously are not. That is enough to make good decisions.
What to do first
Add a source field to your job records and fill it in from today. Sixty days of that produces a better answer than any retrospective analysis.
Getting help with it
Bring your spend by channel and your last sixty days of jobs. That is enough to start.
Frequently asked questions
What do I need to prove marketing ROI?
Spend by channel, enquiries attributed to channels, which enquiries became jobs, and what those jobs were worth. The third is almost always missing and only you can supply it.
Can ROI be worked out retrospectively?
Partly. Matching invoice dates to enquiry dates, using call tracking data, and simply asking recent customers all help. Rough and roughly right beats precise about the wrong thing.
What is the right comparison?
Cost per booked job by channel, measured against what a lead is worth to you. Click-through rates and return on ad spend in isolation both mislead.
Can anyone give me exact attribution?
No, and anyone claiming to is overselling. A customer who saw an ad, then searched, then asked a neighbour is not cleanly assignable. A defensible ranking of channels is enough to decide well.