Revenue Attribution Workflow

Step 1: Paid touches

Every paid touch comes in with the company behind it: LinkedIn campaigns match on company, Google and Meta match through the tags on the landing pages. Each click lands with its campaign, its date and its cost, so cost per meeting can be computed later, per channel, with nobody guessing. Bringing the three ad platforms in through the same door matters more than it sounds. Each of them reports its own version of a conversion, and left alone all three claim the same meeting. Reading them as raw touches, with a date and a company, is what lets the split step later give each one its real share. Cost travels with every touch, so the board later can say what a meeting from LinkedIn cost against a meeting from Meta, and it can say it per campaign. That's the number the budget conversation needs and almost never has, because the platforms report cost per click and your CRM reports meetings and nothing joins them until this step does.

Step 2: Stamp at the door

Every visit to your site gets identified to a company where possible and stamped with where it came from, so a click from a LinkedIn ad and a visit from a blog post carry different labels from the first second. Outbound emails and meetings get stamped the same way, from your sending tool and your calendar, and the stamp is what makes the timeline honest. Company identification on the site is the piece that connects marketing to the deals: an anonymous visit is a number, and a visit from a company that's in your pipeline is a touch on a real timeline. Where the visitor can't be identified, the visit still keeps its source, so channel totals stay honest even when the account can't be named.

Step 3: Make sources agree

Five tools spell the same company five ways, and the merge makes them agree: one company, one timeline, every touch on it in order. This is unglamorous work and it's where every attribution project we've seen fail went wrong, because a timeline with the touches split across three spellings of the same company credits nothing correctly. The merge uses the domain where it can and a fuzzy match where it must, and every merge decision is written down, so when somebody asks why two visits from "Acme Inc" and "acme.io" landed on one company, the answer is on the record. Manual fixes are rare and they stick, because the workflow remembers them the next time the same spelling comes through.

Step 4: Broken tags

Tags break more often than anybody admits: a landing page gets rebuilt and the tracking parameter falls off, or a campaign launches without one, and a month of clicks lands as "direct." The detector finds the gaps by watching for traffic patterns that suddenly lose their source, and it flags them to whoever owns the page, so a broken tag lives a day and not a quarter. Finding the break the day it happens is what saves the quarter's numbers. A tag that's fixed within a day leaves a gap the fill step can close with confidence, and a tag that's broken for two months leaves a hole that nobody can honestly fill, and the whole board gets argued about again. Teams without the detector find out when the board looks wrong, which is usually a quarter later.

Step 5: Close the gaps

Where a source went dark, the workflow fills the gap from what it can see: a spike from a LinkedIn campaign's dates, an email sequence's send log, a meeting on the calendar. Every filled touch is marked as inferred, so the board always shows what was measured and what was reconstructed, and nobody mistakes one for the other. The rule for filling gaps is conservative on purpose: the workflow only infers a touch when there's a clear cause on the calendar, and it never invents one to make a channel look better. Inferred touches carry a lower confidence, and they show up as a separate shade on the board, so the story is honest about how much of it was measured.

Step 6: Split the credit

The touches that preceded each deal share the credit, by a rule your team agrees once: first touch, last touch, or a spread across all of them. We tell clients the rule matters far less than having one rule that everybody uses, because the fight is never about the model, it's about three dashboards each using a different one. The rule is written down once and applied everywhere, and the board shows which rule it used, so anybody who disagrees is arguing with a rule they can read and not with a dashboard they can't. When you change the rule, the whole history recomputes, so you can see what last quarter looks like under first touch and under last touch and decide with the numbers in front of you. The board also shows the touches that got no credit under the current rule but sat on the timeline anyway, because a channel that opens conversations and never gets the last touch is easy to cut by mistake. We've watched a team stop a content programme that a first-touch view would have shown was starting a third of their pipeline.

Step 1: Paid touches

Every paid touch comes in with the company behind it: LinkedIn campaigns match on company, Google and Meta match through the tags on the landing pages. Each click lands with its campaign, its date and its cost, so cost per meeting can be computed later, per channel, with nobody guessing. Bringing the three ad platforms in through the same door matters more than it sounds. Each of them reports its own version of a conversion, and left alone all three claim the same meeting. Reading them as raw touches, with a date and a company, is what lets the split step later give each one its real share. Cost travels with every touch, so the board later can say what a meeting from LinkedIn cost against a meeting from Meta, and it can say it per campaign. That's the number the budget conversation needs and almost never has, because the platforms report cost per click and your CRM reports meetings and nothing joins them until this step does.

Step 2: Stamp at the door

Every visit to your site gets identified to a company where possible and stamped with where it came from, so a click from a LinkedIn ad and a visit from a blog post carry different labels from the first second. Outbound emails and meetings get stamped the same way, from your sending tool and your calendar, and the stamp is what makes the timeline honest. Company identification on the site is the piece that connects marketing to the deals: an anonymous visit is a number, and a visit from a company that's in your pipeline is a touch on a real timeline. Where the visitor can't be identified, the visit still keeps its source, so channel totals stay honest even when the account can't be named.

Step 3: Make sources agree

Five tools spell the same company five ways, and the merge makes them agree: one company, one timeline, every touch on it in order. This is unglamorous work and it's where every attribution project we've seen fail went wrong, because a timeline with the touches split across three spellings of the same company credits nothing correctly. The merge uses the domain where it can and a fuzzy match where it must, and every merge decision is written down, so when somebody asks why two visits from "Acme Inc" and "acme.io" landed on one company, the answer is on the record. Manual fixes are rare and they stick, because the workflow remembers them the next time the same spelling comes through.

Step 4: Broken tags

Tags break more often than anybody admits: a landing page gets rebuilt and the tracking parameter falls off, or a campaign launches without one, and a month of clicks lands as "direct." The detector finds the gaps by watching for traffic patterns that suddenly lose their source, and it flags them to whoever owns the page, so a broken tag lives a day and not a quarter. Finding the break the day it happens is what saves the quarter's numbers. A tag that's fixed within a day leaves a gap the fill step can close with confidence, and a tag that's broken for two months leaves a hole that nobody can honestly fill, and the whole board gets argued about again. Teams without the detector find out when the board looks wrong, which is usually a quarter later.

Step 5: Close the gaps

Where a source went dark, the workflow fills the gap from what it can see: a spike from a LinkedIn campaign's dates, an email sequence's send log, a meeting on the calendar. Every filled touch is marked as inferred, so the board always shows what was measured and what was reconstructed, and nobody mistakes one for the other. The rule for filling gaps is conservative on purpose: the workflow only infers a touch when there's a clear cause on the calendar, and it never invents one to make a channel look better. Inferred touches carry a lower confidence, and they show up as a separate shade on the board, so the story is honest about how much of it was measured.

Step 6: Split the credit

The touches that preceded each deal share the credit, by a rule your team agrees once: first touch, last touch, or a spread across all of them. We tell clients the rule matters far less than having one rule that everybody uses, because the fight is never about the model, it's about three dashboards each using a different one. The rule is written down once and applied everywhere, and the board shows which rule it used, so anybody who disagrees is arguing with a rule they can read and not with a dashboard they can't. When you change the rule, the whole history recomputes, so you can see what last quarter looks like under first touch and under last touch and decide with the numbers in front of you. The board also shows the touches that got no credit under the current rule but sat on the timeline anyway, because a channel that opens conversations and never gets the last touch is easy to cut by mistake. We've watched a team stop a content programme that a first-touch view would have shown was starting a third of their pipeline.

Step 7: Build one true board

One board, one number per channel: touches, meetings, pipeline, revenue and cost. Marketing, sales and leadership read the same board, and the argument about where revenue came from turns into a conversation about where to put next quarter's budget. It refreshes on its own and lives in accounts you own. The board also carries the inferred touches marked as inferred and the gaps that couldn't be filled, because a board that hides its own uncertainty gets trusted for a month and then abandoned. What we've seen is that a slightly humbler board that says "we reconstructed 8% of this" gets used for years. The board is a table and a few charts in the tool your team already uses, and nothing about it is a product you pay for. Because the timelines underneath are yours, in accounts you own, any question the board doesn't answer can be answered by querying them, and after a quarter or two the questions get more interesting than "which channel."

Step 7: Build one true board

One board, one number per channel: touches, meetings, pipeline, revenue and cost. Marketing, sales and leadership read the same board, and the argument about where revenue came from turns into a conversation about where to put next quarter's budget. It refreshes on its own and lives in accounts you own. The board also carries the inferred touches marked as inferred and the gaps that couldn't be filled, because a board that hides its own uncertainty gets trusted for a month and then abandoned. What we've seen is that a slightly humbler board that says "we reconstructed 8% of this" gets used for years. The board is a table and a few charts in the tool your team already uses, and nothing about it is a product you pay for. Because the timelines underneath are yours, in accounts you own, any question the board doesn't answer can be answered by querying them, and after a quarter or two the questions get more interesting than "which channel."

What is a revenue attribution workflow and how we set it up?

Ready to implement this workflow?

Ready to implement this workflow?

Ready to implement this workflow?