Where Did That Lead Come From? Calls, Forms and the Attribution Gap

What you’ll take away

• Attribution is a rule dressed up as a measurement. Credit isn't discovered, it's assigned.
• Google Ads and GA4 disagree for four boring reasons: conversion date, counting rules, cross-device identity, and the attribution model in use.
• The model matters far less than being consistent with it. Changing models mid-year makes your comparisons worthless.
• Tag every link you control with UTMs. Untagged traffic disappears into "direct".
• Dynamic number insertion tracks calls that start on your website — but not calls placed straight from a search result or map listing.
• Offline conversion tracking is what teaches the bidding system to chase real orders instead of cheap form fills.
• ROI is a defensible estimate, not a fact. Judge a team by whether they can name the assumptions underneath the number.
• To test whether ads really work, watch branded search volume — or pause a campaign for four weeks and see what happens to total inquiries.

Transcript

Eden: This is The Visibility Room. Today's episode: where did that lead come from — calls, forms, and the attribution gap. Your host is Ray, who leads paid media and advertising at Mediaforce. Joining him is Erin, who is a sales director at an industrial equipment supplier in Kitchener–Waterloo, and has been in the business twelve years. Here they are.

Ray: Thanks, Eden. Erin, good to have you.

Erin: Thanks, Ray.

Ray: So you brought me a piece of paper before we started, which I loved.

Erin: I did. Okay, so this is one month. Google Ads tells me forty-two conversions. GA4 tells me thirty-one. My CRM has nineteen actual human beings in it. And my sales guy swears half his good stuff came from a referral.

Ray: That's a beautiful piece of paper.

Erin: It's not beautiful, it's infuriating. Which number is right?

Ray: None of them, and all of them. And I know that's the worst possible answer, so let me actually earn it.

Erin: Please.

Ray: Those four numbers are counting four different things. They're not in conflict — they're not even measuring the same event. Google Ads counts ad interactions that led to something. GA4 counts website behaviour. Your CRM counts people your team decided were real. Your sales guy is counting deals. Every one of those is a legitimate number, and if you line them up expecting them to match, you will lose your mind.

What is attribution in marketing?

Erin: Okay, but start further back for me. What is attribution, actually? Because I use the word and I'm not sure I could define it.

Ray: Attribution is the process of assigning credit for a conversion to the marketing that influenced it. That's it. And the key word is assigning — because credit isn't a fact you discover, it's a decision you make. Somebody sees an ad, reads a review, asks a colleague, then searches your name a month later and calls. Attribution is you choosing which of those gets the point.

Erin: So it's a rule, not a measurement.

Ray: It's a rule dressed up as a measurement. That's the single most useful thing to understand about it.

Why don’t Google Ads and GA4 conversion numbers match?

Erin: Alright. So specifically — why don't Google Ads and GA4 match? Because those are both Google products and I do not understand how they can't agree.

Ray: Yeah, welcome. There are four real reasons and they're all boring, which is the good news, because boring means fixable or at least explainable.

Erin: Go.

Ray: One — the date. Google Ads credits a conversion back to the day of the click. GA4 records it on the day it actually happened. So somebody clicks on the twenty-eighth and fills out your form on the second of the next month — Ads puts that in last month, GA4 puts it in this month. Right there, your months don't line up.

Erin: Oh, that's — I've been comparing those side by side for two years.

Ray: Everyone has. Two — counting rules. Ads can count every conversion from one person; GA4 might be set to count them once. Somebody submits your form and then calls you an hour later, that's two conversions in one system and possibly one in the other. Three — identity. Ads can connect a click on somebody's phone to a purchase on their laptop using signed-in Google data. GA4 relies on its own identifiers, and browser privacy changes have chipped away at those for years.

Erin: And four?

Ray: Four is the model. Ads and GA4 can be running different attribution models, so they'd distribute credit differently even from identical data.

Last-click vs data-driven attribution: which should you use?

Erin: That's the word I keep hearing — model. Last click, data-driven. What's the difference and which should I be using?

Ray: Last click gives one hundred per cent of the credit to the final ad interaction before the conversion. It's simple, it's stable, and it systematically under-credits everything that happened earlier — the awareness stuff, the first search, the video. Data-driven attribution spreads credit across the touchpoints based on patterns in your account. It's more realistic, and it's harder to explain to a room.

Erin: So which one?

Ray: Honestly? Whichever one you'll be consistent with. That's not a dodge — the model matters far less than not switching it. If you change models halfway through the year, your before-and-after comparison is garbage and you'll make a bad call off it. Pick one, write down which one it is, and compare like to like.

What is a UTM parameter and do you need them?

Erin: What about everything that isn't Google? We do LinkedIn, we send an email newsletter, and we've got a QR code on our trade show booth.

Ray: Then you need UTMs, and this is the cheapest fix on the entire list.

Erin: What is a UTM?

Ray: A UTM is a short tag you add to the end of a link that tells your analytics where that click came from. So instead of sending people to your homepage, you send them to your homepage with a tag that says: this one came from LinkedIn, from the October campaign. Same page, same experience for the visitor — they don't see anything different — but now the visit arrives labelled instead of anonymous.

Erin: And without it?

Ray: Without it, that traffic mostly lands in a bucket called direct, which is the analytics equivalent of a shrug. Direct is where unlabelled traffic goes to disappear. If your direct number is enormous, that's usually not loyal customers typing your web address from memory — it's untagged links.

Erin: So the trade show QR code—

Ray: Tag it. Tag the newsletter, tag the LinkedIn posts, tag anything you can control the link on. It takes about a minute per link and it retroactively makes months of reporting worth reading.

Erin: Okay. Phone calls. Because in my business, most serious inquiries phone. They don't fill out a form for a forty-thousand-dollar piece of equipment.

Ray: No, they don't. And phone is where most attribution quietly falls apart.

How do you track phone calls from Google Ads?

Erin: So how do you track a phone call from an ad?

Ray: The standard method is dynamic number insertion. You put a small piece of code on your website, and it swaps the phone number a visitor sees depending on how they arrived. So somebody from a paid search ad sees one number, somebody from organic sees another. Same phones ring on your end, but now the call carries a source with it, and it can be pushed into your ad platform as a conversion.

Erin: And that just — works?

Ray: It works well, with two caveats I want to be straight about. First, it only covers calls that start on your website. If somebody sees your ad, doesn't click, and calls the number straight off the search result, or off your Google Business Profile listing, that's tracked differently or sometimes not at all. Second, some platforms mask the caller's real number for privacy, so you get a proxy number, not the person's actual line.

Erin: Which matters because my team wants to call them back.

Ray: Exactly, and that's the practical annoyance nobody warns you about.

How do you track conversions with a long sales cycle?

Erin: Okay, here's my bigger one. Our sales cycle is three to six months. A guy calls in March, goes quiet, comes back in July with a purchase order. How on earth do I connect that back to an ad from March?

Ray: Offline conversion tracking. And this is the piece that separates accounts that actually optimize from accounts that just spend.

Erin: Explain it like I'm going to have to sell it to my IT guy.

Ray: Fair. When someone converts on your site, the ad platform generates a click identifier — think of it as a receipt number. If that receipt number gets captured into your CRM alongside the lead, then months later when that lead becomes a customer, you send the outcome back to the ad platform: this receipt turned into a real order worth this much.

Erin: And then what happens?

Ray: Then the platform's bidding stops chasing form fills and starts chasing the kind of form fill that becomes a purchase order. Which is a completely different thing. Without that loop, you're telling the algorithm that every inquiry is equally valuable, and it will very obediently go find you more of the cheap worthless ones.

Erin: That explains a lot about last spring.

Ray: It explains a lot about most accounts.

Erin: Is there a version of this that doesn't need my IT guy?

Ray: Yes, and I'd argue it's eighty per cent of the value. Tag every lead by outcome — real, junk, unreachable, quoted, won — and once a month look at cost per real lead instead of cost per lead. You can do that in a spreadsheet. It won't feed the algorithm automatically, but it'll tell you and me which campaigns to cut, and that's the decision that actually saves money.

Erin: That I can start Monday.

Ray: That's the one I'd start Monday.

Can you actually calculate marketing ROI?

Erin: So — the big question. Can I ever actually calculate ROI? Because that's what my owner asks me, and I never have a clean answer.

Ray: You can calculate a defensible estimate. You cannot calculate a fact, and anyone who hands you a fact is hiding assumptions from you.

Erin: Walk me through the defensible version.

Ray: Let's use round numbers. Say you spend eight thousand dollars in a month. Your CRM shows twenty real inquiries — not forty-two, twenty, the ones a human confirmed. So four hundred dollars per real inquiry. Of those twenty, eight get to a formal quote. Of those eight, two become orders, at say twenty-two thousand each. That's forty-four thousand in revenue.

Erin: On eight thousand of spend.

Ray: On eight thousand of spend. But don't stop there, because revenue isn't return. Take your gross margin — say thirty per cent — and that forty-four thousand is about thirteen thousand in gross profit. Against eight thousand in ad spend, that's real but it's not the five-to-one story the revenue number implies.

Erin: And my owner would absolutely quote the five-to-one.

Ray: Everybody's owner quotes the five-to-one. And here's the assumption I'd flag out loud: those two orders came from inquiries that probably started three to six months ago, not from this month's eight thousand. So you're dividing this month's cost by last quarter's results.

Erin: So the number's wrong.

Ray: The number's directional. Which is fine, as long as you say so. And the fix is easy — measure it as a cohort. Take the leads that came in during March, and follow those specific leads to whatever they became. Now the spend and the outcome belong to each other.

Erin: That's the myth then, isn't it. That there's one ROI number.

Ray: That's exactly the myth. Attribution is not exact and it was never going to be. It's a best-evidence picture. The goal is directionally reliable and consistently measured — not accurate to the decimal. And the practical test of whether your team is any good is not whether they give you a precise number. It's whether they can tell you which assumptions are underneath it.

Do ads drive word-of-mouth and branded search?

Erin: Okay, then what about the referral thing? Because that's my sales guy's whole argument. "It's all word of mouth, the ads aren't doing anything."

Ray: So this is my favourite one, because he might be right and he also might be completely wrong, and the way you find out is not by arguing about it.

Erin: We've been arguing about it for a year.

Ray: Here's what's actually happening. Somebody sees your ad, doesn't click. Two weeks later they mention your name to a colleague. That colleague searches your company name and calls. Your CRM says referral. Your sales guy says referral. And in a sense it is a referral — it's just a referral your ad seeded.

Erin: So how do I ever prove that?

Ray: One clean way. Watch your branded search volume — how many people are searching your company name specifically. If that number climbs while your ads are running and drops when you pause, your advertising is generating demand that shows up under other labels. You can see that in Search Console and in your ad account for free.

Erin: And the harder way?

Ray: Turn it off. Genuinely. Pick a slower month, pause a campaign for three or four weeks, and watch what happens to total inquiries — not just the ones tagged paid. That's a real test, and it's the only one that answers the incrementality question honestly. It's uncomfortable, and most people won't do it.

Erin: My owner would have a heart attack.

Ray: Most owners would. But it's the difference between believing your ads work and knowing.

Where to start with attribution

Erin: Alright. What do I actually do this week?

Ray: Three things, and none of them need a developer. First — pick one system as your source of truth for decisions. For a business like yours that's the CRM, because it's the only one that knows what a lead became. Everything else is a supporting instrument.

Erin: Ads and GA4 become diagnostics.

Ray: Diagnostics, not scoreboards. Second — start tagging outcomes on every lead. Real, junk, unreachable, quoted, won. Four seconds a lead, and in ninety days it changes every budget decision you make.

Erin: And third?

Ray: Third — write down your attribution model and your reporting window on an actual piece of paper, and don't change either for a year. Most of what looks like a performance swing in this industry is just somebody quietly changing how things get counted.

Erin: That's a very unglamorous list.

Ray: It's a deeply unglamorous list. It's also the one that works. And honestly — the fact that you walked in here with four numbers that don't match means you're already ahead of most people, because most people only ever look at the one that makes them feel best.

Erin: I'll take it.

Ray: Erin, thanks. This was great.

Erin: Thanks, Ray.

Eden: That was The Visibility Room. Thanks for spending some time with us — we'll see you next episode.