How to tell which marketing channel is actually bringing you customers
Each platform reports on the customers it touched, and their totals rarely add up to the customers you actually won. The answer is to start from the customer and work backwards.
Start from the customers, not the channels. List every new customer from your invoices, trace each one to their first inquiry, record how they say they found you, and only then compare the result with what each platform reports. The channels that appear most often in that list, relative to what they cost, are the ones bringing you customers.
Most companies that invest in marketing run several channels at once: search advertising, social media, a website, email, events, partners, and referrals that nobody pays for directly. Each channel comes with its own report, and each report tends to show a healthy return. Add the reported results together and the total is often larger than the number of customers the business actually won.
That is not a sign that anyone is being dishonest. It is how the platforms are built. Understanding why it happens is the first step to getting a reliable answer.
Why every channel looks like it's working
Advertising platforms measure conversions using their own rules. Each decides what counts as a conversion, how long after an ad view or click a conversion can be credited, and whether a view without a click counts at all. A buyer who saw a social ad, later searched for your company name, clicked a search ad and filled in a contact form can be counted by both platforms as their conversion.
Two further gaps make the picture more generous than reality. First, platforms usually count actions such as form fills or calls, not paying customers, because they cannot see your invoices. Second, they cannot see channels they are not part of. A referral, a conversation at an event, or a recommendation in an industry group leaves no trace in an ad account, so it is never credited to anything.
Each platform reports the customers it touched. Only your books show the customers you won.
What each source of evidence can and cannot tell you
| Source | What it shows | Blind spot | Reliability |
|---|---|---|---|
| Ad platform reports | Clicks, form fills and conversions it was involved in | Other channels, and whether the lead ever paid | Overstates |
| Website analytics | Which source sent each visit before a form was filled | Phone calls, offline conversations, blocked tracking | Partial |
| CRM source field | The source recorded when the inquiry was logged | Only as good as the discipline of whoever fills it in | Varies |
| Asking the customer | How the buyer remembers first hearing about you | Memory is imperfect, especially over long sales cycles | Strong signal |
| Invoice trace | Which paying customers came from which first inquiry | Needs the inquiry log to be kept | Most reliable |
A practical method in four steps
1. List your new customers
Take every customer who paid you for the first time in the last twelve months, from your invoicing or accounting system. This is the list that matters. Everything else is evidence about how it was produced.
2. Trace each one to the first inquiry
For each customer, find the first recorded contact: the form submission, the email, the call log, the CRM entry. Note the date and whatever source was captured at the time. Where nothing was captured, mark it as unknown rather than guessing. The size of the unknown group is itself useful information about your measurement.
3. Add what the customer says
If your team asked how the customer found you, add that answer alongside the recorded source. Where the two disagree, the disagreement is often revealing. A customer recorded as "Google search" who says they were referred by a colleague probably searched your name after the referral. The search ad did not create that customer; it collected them.
4. Compare with what each channel cost
Group the customers by channel, then set the full cost of each channel, including fees and the time of the people who run it, against the number of customers it produced. The result will rarely match the platform reports, and it will usually reorder the channels by value.
Demand creation and demand collection
The trace often shows a pattern worth understanding. Some channels create demand: they introduce your company to people who had not heard of it. Referrals, events, partners, content and some social advertising tend to work this way. Other channels collect demand that already exists: branded search ads and retargeting usually work this way, capturing people who already intended to contact you.
Collection channels look excellent in platform reports because they sit close to the moment of conversion. Creation channels look weaker because their effect arrives later and through other routes. A company that cuts its creation channels because they look expensive, and keeps its collection channels because they look efficient, often sees inquiries fall a few months later without understanding why.
How to keep the answer reliable going forward
- Log every inquiry in one place, with the date, the channel it arrived through, and the source the person gives. One shared record is better than a sophisticated system nobody updates.
- Ask the source question early, in the first conversation, with an open answer.
- Reconcile quarterly. Match the customers in your books with the inquiries in your log, and compare with what the platforms claim.
- Judge channels on customers and cost, over a period at least as long as your typical sales cycle.
Knowing which channel works is the core question of the first area of a growth system, attracting customers. It is also one that cannot be answered without the fourth, measuring results. That connection is why we always examine the two together.
Running several channels and unsure which ones pay?
A short call is enough to tell whether your records can answer the question, and what it would take to answer it properly.
Request an introductory callCommon questions
Why do my ad platforms report more conversions than I have customers?
Each platform counts the conversions it was involved in, using its own rules and time windows. When a buyer sees an ad on one platform, searches on another and later fills in a form, more than one platform can claim the same customer. Platforms also count actions such as form fills, which are not the same as paying customers.
Is it worth paying for an attribution tool?
For most companies between two and fifteen million in revenue, not as a first step. The volume of customers is often too small for statistical models to be reliable, and the tool still depends on clean records. A disciplined inquiry log and a source question usually provide more trustworthy answers at a fraction of the cost.
What question should we ask new customers about how they found us?
Keep it open and ask it early: 'How did you first hear about us?' with a free-text answer. Multiple-choice lists tend to steer people towards whichever option is listed first or looks most familiar.
How often should we review which channels are working?
Quarterly is usually enough for channel decisions, using at least two quarters of data. Monthly reviews are useful for managing the channels themselves, but too short to judge them on customers when sales take several weeks to close.