Expansion Works Insight
Marketing attribution often appears more precise than it really is. This article examines the difficulty of determining what actually caused a customer to respond and what business owners should pay attention to instead.
Attribution is the question “which marketing effort actually caused this lead or sale?” Everyone wants a clean answer. Almost no one has one. The reason: customer journeys are messy, but our tracking systems force them into single-source buckets. The bucket that gets credit is usually the last thing the customer touched — not the thing that actually moved them.
This creates a predictable failure mode. Channels that intercept already-interested people look like heroes. Channels that create interest in the first place look useless. So you cut the channels doing the real work and pour money into the ones taking credit for it.
Scenario 1: The brand-name Google search. Someone sees a billboard on the 401. Two days later, they Google “Canada First” and click the top result — which happens to be your own paid ad bidding on your brand name. They fill out the form. Google Ads claims the lead. The billboard gets nothing. But the billboard created the search; the Google ad just collected the toll on traffic that was already coming.
Scenario 2: The website form black hole. A lead fills out the form on your site. Source: “Website Form.” But that tells you nothing about how they got to the website. Did they see a digital ad? Get referred by a current client? Hear you on a podcast? Drive past a billboard? “Website Form” is a destination, not a source. If a large chunk of leads bucket here, you're effectively flying blind on the upstream channels.
Scenario 3: The post-conversation survey. You ask new clients “how did you hear about us?” They say “Google.” Of course they do — that's the verb people use for “the internet.” Someone who saw a billboard, mentioned it to their spouse, then Googled you to find the website will still answer “Google.” Self-reported attribution is directionally useful but rarely accurate.
Scenario 4: The brand campaign that “doesn't work.” Billboards, brand digital ads, sponsorships — these almost never get direct attribution credit because nobody clicks a billboard. But they're often doing the heaviest lifting upstream, making every other channel cheaper and more effective. Kill them based on attribution data and you'll watch your “winning” channels mysteriously get more expensive over six months.
The tell that attribution is broken is when the numbers don't add up. If your reported cost-per-lead from each channel, multiplied by lead volume, doesn't roughly match your actual marketing spend producing actual closed business — something's miscounted. Either channels are taking credit they shouldn't, channels are getting no credit they deserve, or both.
1. Separate “first touch” from “last touch” in your tracking. Last touch tells you who closed the door. First touch tells you who opened it. Both matter. If your CRM only captures one, you're missing half the picture. For the website form specifically, capture UTM parameters on the URL the lead arrived from, plus a “first visit source” cookie that persists across sessions.
2. Add a free-text “what made you decide to reach out today?” field. Different from “how did you hear about us.” This question gets at the trigger — the thing that moved them from awareness to action. The answers will surprise you and they're more honest than dropdown menus.
3. Run holdout tests on individual channels. This is the only way to know a channel's incremental value — the leads that wouldn't have come without it. Pick one channel. Turn it off for 3–4 weeks. Hold everything else steady. Watch what happens to total lead volume, not just that channel's reported leads. Then turn it back on and test a different channel. Start with the cheapest channels first so the cost of being wrong is low. Brand-name Google search is a classic candidate — many businesses discover they're paying for clicks they'd get organically for free.
4. Accept that brand campaigns require a different measurement model. Billboards and brand digital won't show up cleanly in attribution reports, and that's fine. Measure them on aggregate effects: are total leads up? Is cost-per-lead on your performance channels going down? Are people arriving with higher intent? Brand work is a tide that lifts all boats; you measure the tide, not individual boats.
5. Build a routine, not a one-time audit. Attribution drift happens constantly — new channels, changed customer behavior, platform algorithm changes. Pick a monthly cadence to review: total spend vs. total qualified leads, channel-by-channel cost-per-lead trends, and one holdout test per quarter on a different channel.
You won't get perfect attribution. Nobody does. The goal isn't precision — it's enough clarity to make confident allocation decisions. You'll know you're succeeding when:
The math reconciles.
You can defend why each channel gets the budget it gets, with evidence.
You stop having “is this working?” debates and start having “how do we scale this?” debates.
Brand and performance budgets are evaluated separately, on their own terms.
The discipline is sequencing: fix measurement first, then optimize what you can see, then diversify from a position of clarity. Most companies skip step one and wonder why steps two and three keep failing.
Back to Insights
