Marketing Attribution: How to Tell Which Channels Actually Bring You Customers

What marketing attribution is, and why it is hard

Marketing attribution assigns credit for a conversion to the touchpoints that led to it: the ad someone saw, the search they ran, the email they opened. It is hard because buyers touch several channels, many touches cannot be seen, and every tool counts from its own point of view.

Take a real-looking journey. A homeowner sees your Instagram ad on Monday, does nothing, searches your business name on Thursday, clicks your Google Business Profile, and calls. Meta may count a view-through conversion. Google may count the click. Your analytics may credit organic search or the Business Profile link. The call itself might not be tracked anywhere. One customer, three claims, and possibly zero records of the actual call.

Why it matters for budget

Attribution decides where the next dollar goes. If you credit everything to the last click, you will cut the channels that start journeys and overfund the ones that close them, until the closing channels run out of people to close. If you trust every platform's own numbers, you will think marketing produced far more customers than it did.

Why platform numbers never add up

Each ad platform reports conversions it can connect to its own ads, using its own rules and windows. None of them subtract what other channels did. Add the platforms together and the total is almost always higher than the customers in your CRM.

A worked example (illustrative numbers)

Nothing is broken. Meta counted people who saw an ad and later converted after a Google click. Google counted people who clicked a search ad and then called after a friend's recommendation. Both are telling the truth from where they sit. The mistake is adding them up, or judging each one only by its own report.

The fix is a single source of truth for counting leads and customers, usually your CRM, with each record carrying where the lead came from. Platform numbers then become what they are good for: comparing campaigns and ads inside that platform.

The attribution models that still exist

The list of attribution models got much shorter in 2023. Many guides still describe first click, linear, time decay and position-based models as options you can pick in Google's tools. You cannot, anymore.

Google's Google Analytics attribution documentation lists three models for attribution reports: data-driven, paid and organic last click, and Google paid channels last click. It states that first click, linear, time decay and position-based models are no longer available as of November 2023. The Google Ads help page on attribution models says the same models are no longer supported in Google Ads, that data-driven is the default for most conversion actions, and that last click is the other option.

Lookback windows

The model is half the setting; the window is the other half. In GA4's attribution settings, the default lookback window is 30 days for acquisition key events (first visit, first open) and 90 days for all other key events, with shorter options available. Touches before the window get no credit at all.

Our three-source marketing attribution method

We do not try to find one perfect model. We compare three imperfect sources that fail in different ways, and trust the conclusion when they agree. When they disagree, that is where we test.

1. Platform reports (Google Ads, Meta, Microsoft). Biased toward themselves, but the only place to compare ads, keywords and audiences within one platform.
2. Analytics (GA4 with clean tagging). One rulebook across channels for website visits and key events, but blind to calls it cannot see, to offline sales and to views without clicks.
3. CRM and self-reported source. The lead's first tracked source captured on the record, plus the answer to "How did you hear about us?" Tied to real customers and revenue, but self-reports are fuzzy and first touch misses the middle of the journey.

How the three combine

Ask customers how they found you

Self-reported attribution means asking every new lead or customer how they heard about you, and recording the answer in the CRM. It catches what tracking misses: the podcast, the yard sign, the neighbor, the TikTok they watched on someone else's phone.

How to ask so the answers are usable

• Ask on the form and on the phone, with the same options in both places.
• Use a short pick list plus "Other (tell us)". Free text alone produces answers nobody can tally.
• Keep options in the customer's language: "Google search", "Google Maps", "Instagram or Facebook", "Friend or family", "Saw your truck or sign", "AI assistant like ChatGPT".
• Do not make it required on web forms if it costs you leads; make it required in the CRM when staff log a call.
• Review the "Other" answers monthly and add any that keep repeating.

Expect the answers to disagree with your analytics. A customer who clicked a Google ad after seeing your Instagram posts for a month will often say Instagram. Neither source is wrong; together they show the start and the end of the journey.

Close the phone-call gap

For most local and service businesses, phone calls are the largest blind spot in marketing attribution. If a visitor reads your page and then calls, and nothing records where that visitor came from, the call shows up as nothing at all.

Call tracking fixes it by showing a different number depending on the visitor's source, then recording the source with each call. Combined with tracking for forms, you get a record of every lead and its origin, which is the raw material for everything else in this article.

Our platform, UnifiedIQ, is built for this: it tracks every call and form back to its source. For Nusbaum Medical Centers, that came to 1,645 tracked inquiries in the past 12 months, each with its channel attached.

Keep the source on the record

Tracking the call is step one. Step two is carrying the source into the CRM so that when the lead becomes a customer, the revenue can be tied back to the channel. Our Google Ads conversion tracking guide shows how to send closed deals back to Google so bidding learns from them, and fast follow-up matters as much as tracking; see our speed to lead guide.

Test what attribution cannot prove

Attribution describes what happened before a conversion; it cannot prove a channel caused it. For big budget decisions, run a holdout test: turn a channel off or down in one place or period, keep everything else steady, and measure the difference in leads.

A simple holdout test

1. Pick one question. For example: does our retargeting produce customers who would not have come back on their own?
2. Choose the split. Pause it in some service areas and keep it running in comparable ones, or alternate on and off in planned blocks of weeks.
3. Fix the measure in advance: qualified leads and customers in the CRM, not platform conversions.
4. Run long enough to see past normal week-to-week swings. Low-volume businesses need longer tests.
5. Compare and decide. If leads barely move when the channel is off, it was taking credit for customers you would have had anyway.

Retargeting and branded search ads are the usual first candidates, because they reach people already close to buying. Our guide to retargeting ads covers when that spend earns its keep. Holdouts are also how we settle the classic debate in our Facebook ads vs Google ads comparison: not with each platform's report, but with leads in the CRM.

Which source to trust for which decision

Different questions need different attribution sources. Using the wrong one is how businesses cut good channels and keep bad ones.

The SEO row has its own method, covered in our guide to SEO reporting. For the wider set of numbers each of these feeds, see our marketing KPIs scorecard.

Clean data comes before any model

No model can rescue messy inputs. Most marketing attribution problems we inherit are data problems: untagged links, invented channel labels, uncounted calls, leads without a source in the CRM.

The attribution readiness checklist

• Every external link you control is tagged with consistent UTM values. Our UTM parameters guide has the naming convention.
• Unassigned traffic in GA4 is small. If it is not, your mediums do not match the default channel rules.
• Google Ads auto-tagging is on, and Google Ads is linked to GA4.
• Calls are tracked by source, not just counted.
• Forms capture source and campaign into hidden fields that reach the CRM.
• The CRM records which leads became customers and what they were worth.
• Staff ask and log "How did you hear about us?" on every new lead.

Work through that list before arguing about models. Once it is in place, attribution stops being a debate about whose report to believe and becomes a monthly look at customers by source.

Marketing attribution for small businesses: what to do first

A small business does not need multi-touch software to get marketing attribution right. It needs every lead recorded with a source, and a monthly habit of looking at customers by channel.

After two or three months, that table will tell you more than any attribution model. Then pick the one channel you are least sure of and design a holdout test for it. That order of work also feeds straight into a digital marketing strategy review.

If you would rather have it built for you, our analytics and reporting team sets up the tracking, the CRM fields and the monthly view, and our marketing audit shows which of your current channels the evidence supports. More on our measurement work is on the analytics and growth hub, or you can tell us what you are trying to measure.

Marketing attribution questions, answered