You are spending money in four places. A bit on Google, something on your website, a lead service you have not cancelled, and a sponsorship you agreed to in March.
Ask which one is producing work and the honest answer is usually a shrug. Which means you cannot cut the one that is failing or double the one that is working, and you are guessing with real money.
The good news is that useful attribution is much cheaper and much simpler than the industry suggests. The bad news is that it is never as clean as anyone wants.
Why this is genuinely hard
Two problems, and neither goes away with better software.
The first is that people do not travel in straight lines. Somebody sees your van in the neighborhood, remembers your name a month later when the water heater fails, searches for you by name, reads three reviews, and calls. Ask them how they heard about you and they will say Google, because Google is the last thing they touched. Google gets full credit for a job the van created.
The second is timing. Paid ads report results the same week. SEO and content report results in months. Word of mouth reports whenever it feels like it. If you compare those channels after thirty days, ads win automatically, not because they are better but because they are faster to report. Judging a slow channel on a fast channel's timetable is the single most common way a working investment gets cancelled.
What free gets you
Most of what you need costs nothing.
Ask the question. "How did you hear about us" is imperfect for exactly the reason described above, and it is still worth asking on every single call and form. Write the answer down somewhere consistent. It is directionally useful even when it is precisely wrong, and the pattern across a hundred calls tells you more than any individual answer.
Read your Google Business Profile performance panel. It shows how people found the profile, whether through your name or a general search, and how many then called or clicked through. For most local service businesses this is the closest free thing to a revenue report.
Use Google Search Console. It reports the actual queries that put you in front of people and the clicks you got. Not estimates, actual events.
Tag your links. Any link you control, in an email, a social post, a QR code on a door hanger, can carry UTM parameters so your analytics tells you where the visit came from. It costs nothing but a minute with a link builder.
When to pay for call tracking
For a phone-driven business, dedicated tracking numbers are the one paid tool that genuinely changes what you know. A different number on the truck, the door hanger, the Google profile, and the website, all forwarding to your real phone, and suddenly attribution is a fact rather than a memory.
CallRail restructured its plans in 2026, so be careful with older comparison articles. The current lineup on annual billing runs roughly $50 for Lead Tracking, $95 for Lead Tracking Complete, $150 for Lead Conversion, and $195 for Lead Conversion Complete. Every tier includes five local numbers, 250 minutes, and 25 texts. Month-to-month costs about ten percent more.
Watch the meter rather than the sticker. Extra numbers run around three dollars each and local minutes about five cents. A business taking twenty calls a day at five minutes each burns roughly 2,000 minutes a month, which is well past the included allowance and adds up quickly. Cheaper options exist, with PhoneWagon around $29 a month covering basic tracking.
The rule of thumb worth applying: call tracking pays for itself clearly once you are spending a few thousand a month on advertising. Below a thousand or so, the tracking can cost a meaningful fraction of the spend it is measuring, and a notebook plus the intake question may be enough.
The number to actually watch
Not leads. Not cost per lead. Cost per booked job, and what that job is worth.
A channel producing thirty cheap leads that close at three percent is worse than one producing eight expensive leads that close at twenty five percent, and cost per lead reports the opposite. This is the single most common way a lead vendor looks good on paper while quietly losing you money.
So the ledger has four columns and you can keep it in a spreadsheet: where it came from, whether it became a job, what the job was worth, and what the channel cost that month. Ninety days of that beats any dashboard.
Give each channel its own clock
Before you judge anything, decide up front how long it gets.
Paid search can be assessed in weeks. A direct mail drop needs at least three drops before the result means anything. Content and local SEO work on a horizon of several months. A sponsorship is usually brand exposure that will never attribute cleanly and should be judged on whether you want the relationship, not on tracked calls.
Write the review date down when you start. The alternative is deciding in a slow week that everything is broken.
The trap on the other side
There is a failure mode in the opposite direction, which is over-engineering this.
If you are spending four hundred dollars a month across everything, a hundred and fifty dollar attribution stack is absurd. The measurement should never cost a meaningful share of the thing being measured. At small spend, the intake question plus one tracked number on your highest-cost channel plus a spreadsheet gets you eighty percent of the value for effectively nothing.
Precision is worth buying in proportion to what is at stake.
Where to start this week
Put the question on every call and write the answers in one place. Open your Google Business Profile performance panel and look at it properly, probably for the first time. Then pick the single channel you spend the most on, and give that one a tracked number.
That is an afternoon of work. In ninety days it will tell you which of those four line items to cut, which is very likely worth more than the four line items combined.