Marketing Attribution: How to Tell Which Channels Actually Drive Revenue

Ask ten business owners which marketing channel brings in their best customers, and you’ll get ten confident answers. Pull the actual data behind those answers, and a good number of them fall apart. The channel that “always works” turns out to be the one that gets credit for closing deals that started somewhere else entirely — and the channel quietly doing the heavy lifting is the one on the chopping block at the next budget meeting.

That gap between what you believe about your marketing and what is actually true is what marketing attribution is meant to close. It is not a piece of software you buy, and it is not a dashboard that hands you the answer. It is a discipline: tagging your traffic consistently, capturing where every lead came from at the moment they arrive, and testing your assumptions instead of trusting the default report. Below is how we approach it for small and mid-sized businesses, including where the data will lie to you and what to do about it.

What Marketing Attribution Actually Means

Attribution is the practice of assigning credit for a sale to the marketing touchpoints that contributed to it. In a simple world, someone searches for what you sell, clicks your ad, and buys. One touchpoint, one sale, easy math.

Real buying journeys are messier. A prospect hears your name from a colleague, looks you up on their phone, reads a blog post, forgets about you for three weeks, sees a retargeting ad, searches your business name directly, and finally calls the number on your website from a desk phone. That is six touchpoints across two devices and one channel — word of mouth — that leaves no digital trace at all. Every analytics platform on earth has to make a judgment call about who gets the credit, and different platforms make different calls. That is why your ad platform, your analytics tool, and your CRM will never agree on conversion counts, and why chasing perfect agreement is a waste of a week.

The goal of good marketing attribution is not a perfect ledger. It is a directionally honest picture that is good enough to move budget with confidence.

The Attribution Models, in Plain English

An attribution model is simply a rule for splitting credit across touchpoints. These are the ones you’ll encounter:

Model	How it assigns credit	Best used when
Last click	100% to the final touch before conversion	Short, simple sales cycles; a quick sanity check
First click	100% to the first touch	You’re trying to understand what creates demand
Linear	Split evenly across every touch	Long cycles where many touches genuinely matter
Time decay	More credit to touches closer to the sale	Cycles where late-stage nurturing does the closing
Position-based	Weighted to the first and last touch, rest split	You care about both discovery and closing
Data-driven	Algorithmically, based on observed patterns	You have high conversion volume to learn from

Most platforms now default to some form of data-driven or last-click model. Here is the part people miss: last click systematically overvalues channels that catch people who were already going to buy. Branded search is the classic example. When someone Googles your company name and clicks your ad, that ad gets full credit — but that person already knew who you were. Something else made them know. Kill the channel that created the awareness, and branded search quietly dries up a quarter later.

The channel that closes the sale is rarely the channel that created the customer. Budget decisions made on last-click data alone tend to defund demand and reward demand capture.

Which model should you pick?

For most small businesses, we suggest looking at two models side by side rather than crowning one: a first-touch view to see what creates awareness, and a last-touch view to see what closes. When both views agree that a channel is performing, you can act with real confidence. When they disagree sharply, that’s the signal to dig in — not a data error to be resolved by picking whichever number you like better.

Build the Plumbing Before You Buy the Tool

Attribution software cannot fix inputs that were never captured. Before spending on a platform, get these five fundamentals in place. They cost very little and they solve the majority of the problem.

1. Define what actually counts as a conversion

A newsletter signup and a booked consultation are not the same event, and lumping them together makes every downstream number meaningless. List your conversion events, rank them by how close they sit to revenue, and track them separately. Where you can, attach a value — even a rough internal estimate of what an average qualified lead is worth to you — so you can compare channels on money rather than on raw lead counts.

2. Tag every link you control with UTMs

UTM parameters are the tags appended to your links that tell analytics where a visitor came from. They are free, and they are the single highest-leverage thing most businesses are doing badly. The rules are simple and the discipline is everything:

• Pick a convention and write it down — lowercase everything, use hyphens, never spaces.
• Keep source, medium, and campaign meaningful and consistent. facebook and Facebook and fb will appear as three separate channels forever.
• Tag everything you own: email campaigns, social posts, QR codes on print, partner links, review-site profiles, email signatures.
• Never UTM-tag internal links between pages on your own site. It resets the visitor’s original source and erases the answer you were looking for.
• Keep a shared spreadsheet of every tagged link so the whole team builds them the same way.

3. Capture lead source in your CRM at the moment of entry

This is where attribution stops being an analytics exercise and starts being a revenue one. Analytics tells you a form was submitted; only your CRM knows whether that lead became a $400 job or a $40,000 contract. Pass the original source, medium, and campaign into hidden fields on your forms so the data lands on the contact record automatically, and store both the first-touch and last-touch source. Do not rely on a salesperson typing it in later — it will be blank half the time and wrong a fair share of the rest. If your CRM is not set up to receive this cleanly, that’s a plumbing problem worth fixing first; we walked through the broader version of it in our guide to rolling out a CRM your team will actually use.

4. Track phone calls, not just forms

For service businesses, contractors, medical practices, and anyone with a considered purchase, the phone is often the primary conversion — and it is invisible in standard analytics. Call tracking assigns a dynamic phone number based on how the visitor arrived, so a call from an ad click and a call from an organic search show up as different sources. If a meaningful share of your revenue arrives by phone or text and you aren’t tracking it, your attribution isn’t 80% complete, it’s structurally broken.

5. Just ask people

The most underrated attribution tool is a single optional field on your intake form or your first sales call: How did you hear about us? Self-reported attribution is imprecise — people misremember, and they’ll say “Google” when they mean a podcast that made them Google you. But it is the only method that captures the untrackable: referrals, word of mouth, a conversation at a trade show, a billboard. Use it as a cross-check against your digital data, not as a replacement. When self-reported answers and platform data disagree consistently, the truth is usually somewhere in between and always worth investigating.

Where the Data Will Mislead You

Even a well-built setup has known blind spots. Knowing them keeps you honest:

• Privacy changes and cookie restrictions. Browser tracking protections, consent requirements, and shorter cookie lifetimes all shrink how much of the journey you can observe. Long sales cycles are hit hardest, because the window may close before the deal does.
• Dark social and offline word of mouth. A link pasted into a text thread or a private group arrives as direct traffic with no referrer. “Direct” is not a channel; it’s a bucket of things you failed to tag plus things you can’t tag.
• Cross-device journeys. Researched on a phone, purchased on a laptop — often two separate users as far as your analytics is concerned, unless a login or an email address ties them together.
• Platform self-reporting. Ad platforms grade their own homework. Their attribution windows are generous by design, and view-through conversions can credit an ad nobody consciously saw. Compare their numbers against your CRM, and treat the CRM as the tiebreaker.
• Small numbers. If a channel produced eleven leads last month, its cost-per-lead is noise, not a trend. Judge low-volume channels over quarters, not weeks.

The Test That Beats Any Model: Incrementality

Attribution models answer “who got credit?” The better question is “what would have happened if we hadn’t run this at all?” That’s incrementality, and you don’t need enterprise tooling to test it.

Pause a channel entirely for a defined period — long enough to cover your typical sales cycle — and watch total lead volume and revenue, not just that channel’s numbers. If overall results hold steady, that channel was mostly capturing demand you’d have gotten anyway. If total volume drops, you just proved its value more convincingly than any dashboard could. Businesses operating in multiple areas can run the cleaner version: keep spending in one region and pause in a comparable one, then compare.

A holdout test is uncomfortable, cheap, and more truthful than any attribution model. It’s the only method that measures what your marketing caused rather than what it merely accompanied.

Run these deliberately and rarely — one channel at a time, documented in advance, with the success criteria written down before you start so you can’t rationalize the outcome afterward.

Turning Attribution Into Decisions

Data that doesn’t change behavior is just expensive decoration. Put a rhythm around it:

1. Monthly: review cost per qualified lead and cost per customer by channel, using CRM data rather than platform-reported conversions.
2. Quarterly: compare first-touch and last-touch views for the same period and look for channels the two disagree about. Investigate the gaps.
3. Quarterly: shift budget in increments, not lurches. Move 10–20% toward what’s working and watch a full cycle before moving more.
4. Annually: run at least one honest incrementality test on your largest line item.

Keep this on the same report your team already reads — attribution belongs on your standing scorecard, not in a separate file nobody opens. Our guide to building a small business KPI dashboard covers how to structure that reporting so it drives weekly decisions.

One last thing worth saying plainly: attribution tells you which door people came through, not whether the room they entered was any good. If every channel underperforms at roughly the same rate, the problem is usually not channel mix — it’s what happens after the click. Before you spend another quarter reallocating budget between sources, make sure the landing experience is pulling its weight. That’s the ground covered in our post on turning the traffic you already have into customers.

Common Mistakes We See

• Judging every channel by last click, then wondering why awareness spending “never works.”
• Inconsistent UTM tagging that fragments one channel into six.
• Counting leads instead of revenue, so the channel producing volumes of unqualified inquiries looks like the winner.
• Untracked phone calls in a business that runs on phone calls.
• Changing budget every two weeks based on samples too small to mean anything.
• Buying an attribution platform before fixing the CRM fields it needs to read from.

Start Where You Are

You do not need a perfect system to make better decisions than you’re making today. Standardize your UTMs this week. Add a “how did you hear about us” field to your intake form. Get lead source writing automatically into your CRM. Those three moves alone will tell you more than most businesses ever learn about their own funnel — and they set the foundation for everything more sophisticated that comes later.

If you’d like a second set of eyes on your setup, we’re happy to help. Frozen Crow builds tracking, reporting, and marketing programs for small and mid-sized businesses across Orange County and beyond — and we’ll tell you honestly what your data can and can’t support. Reach out at frozencrow.com for a free, no-obligation consultation.