How Do I Know Which Marketing Channel Actually Brought In a Case?

This sounds like a data question. It is mostly a definition question.

A prospective client sees a billboard on their commute for six months. A friend mentions your firm. They search your name, click a paid ad because it sits at the top, land on your site, leave, come back two weeks later through a search result, and call.

Which channel brought in that case?

Any answer you give is a choice about how to assign credit, not a discovery of the truth. Once a firm accepts that, the problem becomes manageable.

Two sources of evidence, both partial

What the client tells you. Intake asks how they heard about the firm. This is the only method that can see offline entirely — the referral from a former client, the billboard, the sponsorship. It is also unreliable, because people genuinely do not remember. The most common answer at most firms is some version of “I found you online,” which is often the last step rather than the reason.

What the digital trail shows. Call tracking numbers, form source fields, click identifiers. This is precise about the final interaction and blind to everything that happened offline. It systematically over-credits branded search, because someone who already knows your name has to type it somewhere.

Neither is sufficient. Together they are usually enough.

Capture both, and record the disagreement

The practical move is to store both and stop treating them as competing answers.

When intake records a self-reported source and the system records a digital source, keep both fields. When they agree, confidence is high. When they disagree, you have learned something specific: an offline channel probably created the demand and a digital channel captured it.

Firms that force a single source field lose that signal entirely, and it is one of the more valuable things in the dataset. A firm that sees “client said radio, system said branded search” repeatedly is watching its radio spend work, even though no click will ever prove it.

The rule has to be written down

Where the evidence conflicts, something has to decide. That something is a rule, and it needs to exist on paper before anyone analyzes anything.

The common approaches: credit the last interaction, credit the first, split credit across everything that touched the case, or weight the first and last more heavily than the middle.

None of these is correct in the abstract. Last-touch flatters channels that harvest existing demand. First-touch flatters channels that create awareness and undercredits the ones that close. Splitting treats a passing impression as equal to a deliberate click.

What matters far more than which you pick is that it is documented and left alone. A firm that changes its attribution approach mid-year cannot say whether anything improved, because the measurement moved at the same time as the thing being measured. That sounds obvious and it happens constantly, usually because a new vendor prefers a model that shows their channel favorably.

Pick a rule, write it down, and keep it for at least four quarters. The consistency is worth more than the sophistication.

The category that quietly ruins everything

Before refining attribution, check how many of your signed cases have no recorded source at all.

If that number is high, nothing downstream is reliable — and the missing cases are not a random sample. They skew heavily toward phone calls, walk-ins and referrals, because those are the paths where nobody clicked anything and intake was busy.

So the channels that generate calls get systematically under-credited against the channels that generate form fills. A firm acting on that data will shift budget toward form-fill channels and conclude it made an evidence-based decision.

Fixing this is unglamorous and it is the highest-value thing on the list: make the source field mandatory at intake, make sure every inbound route is tracked, and audit how many cases close with it blank.

What “good enough” looks like

You do not need certainty. You need enough confidence to move money.

A firm that knows roughly what share of cases came from each channel, knows how that share moved over a year, and knows which channels its clients mention unprompted has enough to act on. It can reallocate budget, question a vendor, or justify a new channel.

A firm with a precise-looking number derived from partial data is in a worse position, because the precision invites confidence that the underlying data does not support.

Where to start

Check what share of your signed cases carry a source. If it is low, fix that first and wait a quarter before doing anything clever.

Add a self-reported source question at intake if there is not one, and keep it alongside the digital source rather than replacing it.

Then write down your attribution rule, however simple, and put a date on it. That one page is what lets you compare next year to this one — which is the only comparison that reliably tells you anything.