If you have ever sat in a budget review trying to explain which marketing channels actually contributed to revenue, you probably know how quickly the conversation gets messy.
Marketing has one report. Sales has another. The ad platforms have their own numbers. And somewhere in the middle, someone asks the question that actually matters:
Which of these channels created revenue?
This is usually where teams start debating attribution models.
First touch? Last touch? Multi touch?
But in the revenue systems we review at Sention, where the buyer is a company and not one person, the problem often starts much earlier.
The original source was captured. It just did not survive long enough to reach the closed deal.
We call this origin decay.
What Origin Decay Actually Means
Origin decay is the progressive loss of the information that tells us where a prospect originally came from.
A lead might enter the CRM correctly identified as LinkedIn Ads, Organic Search, Referral, Webinar, Outbound, or another source.
Then the record moves.
The lead becomes a contact.
A company or Account record is created.
An opportunity is opened.
Several people from the same company become involved.
The deal eventually reaches Closed Won.
Somewhere between those steps, the original source disappears, gets overwritten, or simply never gets copied to the next record.
By the time we want to connect Marketing to revenue, the CRM knows how much the customer bought, but it no longer knows how the relationship started.
That is not primarily a marketing attribution model problem.
It is a data continuity problem.
When the Buyer Is a Company, Attribution Has to Survive the Account Record Too
This matters even more when the buyer is a company and not an individual, because we are not really selling to individual leads.
We are selling to companies.
When we capture the origin of a lead or contact, that information cannot stay only on the person record. When the CRM creates or identifies the company associated with that person, the relevant attribution data also needs to be carried into the Account record.
This is one of the places where we see attribution break most often.
The lead comes in with a clear source.
The Contact may still have it.
But the Account is created with no origin at all.
Later, Sales creates an Opportunity from that Account, and now the deal starts its life without any reliable information about where the relationship originally came from.
The CRM has not technically failed.
Every record exists.
The problem is that the information never made the full journey:
The information never made the full journey
Lead or Contact
(carries the origin)
→
Account
(no origin)
→
Opportunity
(no origin)
→
Closed Won
(no origin)
When the buyer is a company, that chain matters.
If we lose the origin at the company level, proving Marketing’s contribution later becomes much harder, especially when several contacts are involved in the same buying process.
Where Marketing Attribution Usually Breaks
There are four points I would check first.
1. The click reaches the website, but the CRM never receives the data
Campaign parameters can exist perfectly well in the URL.
That does not mean they exist in the CRM.
The form, tracking layer, integration, or automation still has to write those values into actual fields.
If that does not happen, the record is created without an origin.
Everything downstream inherits the blank.
No attribution model can recover information that was never stored in the first place.
2. The person has an origin, but the company does not
This is especially important when several people from the same company are involved.
Imagine someone downloads a report through a LinkedIn campaign.
The Contact is created with:
First Origin:
LinkedIn Paid
So far, so good.
The CRM then creates or links that person to an Account.
If the Account receives no attribution information, Sales may later create an Opportunity from a company record that has no idea where the relationship began.
Now the data exists somewhere in the CRM, but not where the revenue process actually continues.
That is enough to break reporting.
When we design attribution for a company-level buyer, we need explicit rules for what information moves from the person to the company and, later, from the company into the Opportunity.
3. A new interaction overwrites the original source
A prospect discovers the company through a paid LinkedIn campaign in March.
In July, the same person returns through Google.
If the CRM only has one Source field and updates it with the latest interaction, the record now says:
A new interaction overwrites the original source
no value left in the record
Organic Search
Nothing crashes.
There is no error notification.
But the campaign that originally created the relationship has disappeared from the data.
This is why First Origin and Last Origin should not compete for the same field.
They answer different questions.
4. Revenue and origin end up on different records
This gets more complicated when several people from the same company are involved.
One contact may have come from an event.
Another from outbound.
A third may have discovered us through Google.
The Opportunity belongs to the Account.
The revenue belongs to the Opportunity.
So which origin should count?
There is no universal answer.
But there has to be a rule.
Without one, the CRM often ends up with the worst possible outcome: a large Closed Won deal and a Source field that says Unknown.
Fix the Data Before Choosing the Model
Once origin information survives the full journey, then the attribution model becomes useful.
Different models answer different questions.
| Model | What it helps answer |
|---|---|
| First Touch | Which channel originally created the relationship? |
| Last Touch | Which measurable channel was present before the opportunity progressed or closed? |
| Multi Touch | Which recorded interactions participated across the buying journey? |
| Self Reported | Which source does the buyer themselves believe introduced them to us? |
The mistake is starting here.
A more sophisticated model does not repair incomplete data.
It only distributes incomplete evidence in a more sophisticated way.
The sequence should be:
Preserve the data first. Choose the model second.
First Origin and Last Origin Should Be Separate
A single Source field forces the CRM to choose between preserving history and showing the latest interaction.
We need both.
First Origin
First Origin is written when the relationship is first created and should not be overwritten.
It answers:
What channel created this relationship?
When the buyer is a company, we also need a rule for how that information is inherited or referenced when the Account is created.
Last Origin
Last Origin changes when a new measurable interaction occurs.
It answers:
What channel most recently brought this buyer back?
These are not competing versions of the truth.
They describe two different moments in the same buying journey.
Two different moments in the same buying journey
01
02
03
First Origin
Last Origin
The CRM and the Ad Platform Are Not Supposed to Match Perfectly
Another common problem is trying to make Google Ads, LinkedIn, web analytics, and the CRM report the same number.
They are not measuring the same thing.
An advertising platform measures conversions inside its attribution window.
Web analytics measures sessions and events.
The CRM measures Accounts, Opportunities, and revenue.
Imagine LinkedIn generates a form submission in February.
The opportunity is created in April.
The deal closes in July for $85,000.
LinkedIn may correctly report the February conversion.
The CRM correctly reports $85,000 in July.
Those numbers are not contradictory.
They describe different parts of the journey.
They describe different parts of the journey
LinkedIn generates a form submission
February
The opportunity is created
April
$85,000
July
For practical reporting, we use platform data to understand demand generation and CRM data to understand revenue contribution.
For closed-won attribution, the CRM has to become the source of truth because that is where the final commercial outcome exists.
A Source Field Is Not Enough
Even when a Source field exists, the value may still be too vague to support a budget decision.
Take:
Those are very different acquisition mechanisms
Source:
What does that mean?
LinkedIn Ads?
Organic content?
A direct message from Sales?
A referral that happened inside LinkedIn?
Something entered manually three weeks later because someone vaguely remembered where the lead came from?
Those are very different acquisition mechanisms.
A defensible origin record should tell us more than the channel name.
At minimum, we want to know:
- the origin;
- when it was captured;
- how it was captured;
- whether it was tracked or self reported;
- whether it can be overwritten;
- how it should propagate between Contact, Account, and Opportunity;
- who owns the quality of that information.
Good attribution is not just about having a value.
It is about knowing why we trust it.
Ask the Buyer Too
Tracking will never see everything.
Some opportunities start with a referral.
A private conversation.
A podcast.
An event.
A recommendation between two people who never clicked a trackable link.
That is why we like combining tracked attribution with self-reported attribution.
A simple optional question such as:
How did you hear about us?
can recover information no analytics platform can provide.
Sales can then confirm the answer during discovery and normalize it inside the CRM.
The purpose is not to replace tracking.
It is to complement it.
And when the buyer’s answer differs from the tracked source, that discrepancy can be useful.
It often reveals channels that influence demand without generating the final measurable click.
Attribution Comes Before AI
Predictive lead scoring, automated media buying, and AI-assisted revenue systems all depend on the same underlying data.
If origin data disappears between the Contact, Account, and Opportunity, AI does not repair the problem.
It accelerates decisions based on incomplete evidence.
We can get a faster answer.
We can even get a very confident answer.
But confidence is not the same as traceability.
The order should remain:
Structured data
traceability
automation
Four Numbers Marketing Should Be Able to Defend
A useful Marketing and Sales review does not need twenty attribution charts.
I would rather have four numbers we can actually explain.
1. Closed Revenue by First Origin
How much Closed Won revenue can we connect to the channel that originally created the relationship?
2. Pipeline Created by First Origin
Which channels are creating real sales opportunities?
Not leads.
Opportunities.
3. Cost per Opportunity Created
Cost per Lead tells us how efficiently a channel fills the top of the funnel.
Cost per Opportunity tells us whether those leads become something Sales can actually work.
A channel can produce very cheap leads and very expensive opportunities.
When those two rankings disagree, I care more about the second one.
4. Untraceable Revenue
How much Closed Won revenue cannot be connected to a reliable origin?
This number should stay visible.
Suppose the quarter looks like this:
Untraceable Revenue
Closed Won Revenue
Traceable Revenue
Untraceable Revenue
I would rather show the $90,000 gap than distribute it across channels using a formula the underlying data cannot defend.
That number gives us something concrete to fix.
The Closed-Won Attribution Test
There is a simple way to see how strong your attribution actually is.
Open every Closed Won opportunity from the last completed quarter.
For each deal, answer one question:
Where did this opportunity originally come from?
But use only information stored inside your systems.
Do not ask the Sales Rep.
Do not search old emails.
Do not reconstruct the history from calendar notes.
If the CRM can answer the question, you have evidence.
If a person has to remember the answer, you do not have reliable attribution yet.
Now total the revenue from the deals you were able to trace.
If you closed $600,000 last quarter but can only demonstrate the origin of $410,000, the attribution problem is not theoretical.
You have $190,000 of revenue whose origin your current system cannot defend.
That is the real starting point for improving closed-won attribution.
How Do You Prove Marketing Contribution to Revenue?
You do not start with a more advanced attribution dashboard.
You start by making sure origin information survives the revenue process.
Capture it when the relationship begins.
Preserve First Origin.
Update Last Origin separately.
Carry the relevant attribution information from the Contact into the Account.
Make sure the Opportunity inherits or reliably references it.
Add self-reported origin for the interactions tracking cannot see.
And keep untraceable revenue visible instead of pretending you know where it came from.
Once that chain is in place, marketing attribution becomes much less subjective.
Marketing, Sales, and leadership can finally look at the same revenue and understand where the evidence came from.
See Where Your Revenue Engine Loses the Connection
Origin decay rarely happens by itself.
When data stops surviving one handoff, we often find other problems nearby: incorrect ownership, inconsistent pipeline stages, broken Account relationships, incomplete forecasting, or automations working with only part of the record.
The Growth Engine Report maps how Demand, Pipeline, and Revenue connect across your current setup and shows where those connections break.
