Contents8
A customer signs. They had read an article found on Google, received three newsletters, attended a webinar, clicked on an ad, then called a salesperson. Google Ads counts the sale, so does the email tool, and LinkedIn too; added together, the sales reported by each tool exceed those the accounts record. To decide where next year’s budget goes, you need a rule that shares the credit: that is marketing attribution.
This guide explains attribution models without jargon, then the method that holds in B2B, for teams of 20 to 500 people. It extends our guide to marketing automation for SMEs.
In short
Each platform counts the sales it saw; added together, they count too many.
An attribution model is a rule that shares the credit for a sale between the channels that preceded it.
Google retired four of its models in 2023; data-driven and last click remain.
In B2B, the sale is signed away from the website: attribution goes through the CRM, which ties the contact’s source to the sale.
Three sources cross-check each other: the source recorded by the form, the customer’s answer to “how did you hear about us?”, and the sale in the CRM.
Why every tool claims the sale
Each advertising or email platform sees part of the journey: its clicks, its opens, its visits. When a conversion follows a contact it had, it counts it. None of them sees the others. Anyone who has added up the dashboards knows the result: the sum of reported conversions exceeds the number of real sales, and every channel looks profitable on its own.
It is not deception: each tool answers the question it is asked (“did I play a part?”), not the company’s question (“where should the next pound or euro go?”). Answering the second takes a viewpoint that sees every channel, and that viewpoint is yours, not a platform’s.
Attribution models, in plain words
An attribution model is a sharing rule. The most common:
- Last click: all the credit to the last channel before the conversion. Simple, it favours what closes (a search for your name, a reminder email) and ignores what made the company known.
- First click: all of it to the first contact. The opposite: it favours discovery.
- Linear: credit shared equally between all contacts.
- Time decay: more weight to contacts close to the sale.
- Position-based: more weight to the first and last contacts, the rest shared.
- Data-driven: a platform algorithm estimates each contact’s contribution from the journeys it observed.
At Google, the choice has narrowed. The company announced that first click, linear, time decay and position-based models were going away from Google Ads and Google Analytics 4, and that less than 3% of Google Ads web conversions were attributed using those models.
- June 2023
No longer possible to select these four models for a conversion action that did not already use one.
- September 2023
Actions still using them switch to data-driven attribution, or to last click for those who prefer it.
The data-driven model has real value when volumes are high. But it remains a platform’s model, calculated on what that platform sees: it says nothing about the trade show, the salesperson’s call or a customer’s recommendation.
What B2B adds: the sale is signed elsewhere
In B2B, the website almost never sells on its own. It receives an enquiry; the sale is signed weeks later, after meetings, a quote, sometimes a negotiation. An advertising platform that only sees the enquiry optimises for enquiries, not sales. Hence two moves: record the source of each contact when it arrives, and send the signed sale from the CRM back to the platforms, as described in our article on offline conversions between Google Ads and the CRM.
From the field
A client learned it the hard way
On Palmora Property, a property portal we run, the audit of the existing tracking showed that the function meant to send the measurements did not exist: 100% of measurements were thrown away, and the tag container held one tag and no events. Since then, every form on the site is saved to a database before any email is sent, then creates the contact and the deal in the CRM.
The method that holds: three sources that cross-check
Tying a contact to its sale
- The recorded source
With each form, the CRM receives the campaign parameters, the referring site, the landing page and the language.
- The declared source
A “how did you hear about us?” field, free text or a short list.
- The linked sale
The signed deal is attached to the contact, and so to both its sources.
- The review
Each month, sales and their value per channel, compared with what each channel cost.
Campaign parameters are the UTM tags added to the links you share. Google defines them: utm_source identifies where the traffic comes from (a newsletter, for example), utm_medium the type of channel (email, for example), utm_campaign the campaign or offer. A written naming convention, applied by everyone, keeps the same channel from showing up under five names.
The declared source fills in what measurement cannot see. A customer who writes “a colleague told me about you” or “I saw the founder’s video on LinkedIn” provides information no tool would have captured. The two sources do not always agree, and that is useful: the gap shows what your tools miss.
What attribution will never tell you
Part of the journey stays invisible: a post read without a click, a peer recommendation, a conference, a private conversation. No model measures it, and chasing precision to the decimal wastes time.
Key takeaway
An order of magnitude is enough to decide
Attribution is there to move budget, not to hand out medals. Knowing that one channel brings half the signed sales and another almost none is enough to decide; the third decimal place will not change anything.
Awareness channels often look weak in a last-click model: posting from the leader’s LinkedIn profile, the newsletter, case studies. The declared source and the review with sales correct this bias: they are the ones who hear “I’ve been following you for a year”.
Deciding with attribution
Once the three sources are in place, the monthly review answers simple questions. Which channel brings contacts who sign, not just contacts? What does a sale cost per channel, once everything is counted (the method is in our article on customer acquisition cost)? Which segments respond, as our guide to segmentation shows? Automated marketing reporting pulls these numbers every week, and the marketing plan turns them into budget.
Each channel has its own markers: the webinar is measured in sales conversations, content offered for an address in qualified contacts, email campaigns in replies, SEO in a new market in enquiries per language. Attribution brings them back to one unit: the signed sale.
It also assumes the contact reaches the right salesperson at the right time: that is the marketing-to-sales handoff rule. And when AI produces part of the content, attribution tells you which pieces pay off, provided they are reviewed, sourced and signed; what AI changes for agencies is described in AI and digital marketing.
Where to start
First check that every form on your website records the source in the CRM: send yourself a test enquiry with campaign parameters, and look at what arrives. Add the “how did you hear about us?” field. Write the campaign naming convention. In three months, the first review per channel will have real sales to count.
For attribution that runs from the form to the signed sale in the CRM, our digital marketing offer ties each contact to its source.
Frequently asked questions
Which attribution model should a B2B company choose?
In the platforms, data-driven if your volumes allow it, otherwise last click, knowing what it ignores. For your own decisions, the most useful thing is not a platform model but matching, in the CRM, the recorded source, the source declared by the customer and the signed sale, reviewed every month.
Why don’t the Google Ads and CRM numbers match?
Because they do not count the same thing. Google Ads counts the conversions it saw after a click on its ads, often enquiries; the CRM counts signed sales, whatever the channel. An enquiry may never become a sale, and a sale may have started elsewhere. Sending signed sales back to Google Ads brings the two closer together.
Should you ask customers how they heard about you?
Yes. The question costs one form field or one sentence in a meeting, and it captures what measurement cannot see: recommendations, trade shows, posts read without a click. The answer is imperfect, but the gap between declared and measured source shows where your tools are blind. Record it in the CRM, not in a notebook.
Are UTM tags enough for attribution?
They are necessary, not sufficient. They say where a visit comes from, provided every shared link carries them, with a naming convention that is kept. The form must still pass them to the CRM, the sale must be tied to the contact, and the declared source must fill in what no link captures.
Sources
- Google Ads, four attribution models going away (new tab)support.google.com
- Google Analytics, collect campaign data with custom URLs (new tab)support.google.com
Jim · founder of Brain Plus AI
Also in digital marketing
Practical guide8 min
Automated marketing reporting: track what pays off
A useful marketing report does not copy out figures: every week it says what each channel earned, straight from the tools and unedited. The figures to track, how to connect Analytics, Search Console and your CRM, what AI can comment on, and the data pitfalls to know before you start.
Practical guide7 min
Marketing to sales: the handoff rule
Marketing says it sends contacts nobody calls back, sales says the contacts are not ready. A rule written together settles it: who goes to sales, when, with what information, and what comes back to marketing.
Practical guide8 min
Marketing plan for a 50-person company: what goes in it
A useful marketing plan fits in a few pages: where your customers come from, who you target, what marketing must bring to sales, two or three channels, a budget, a calendar and a review every quarter. What it contains, and what it lets you turn down.