LoginSchedule demoLoginSchedule demoAttribution has two halves: who engaged, and what happened after the click. Most broken attribution is last-mile plumbing in your own stack, not the advocacy data.
The first attribution meeting I run with every program has the same moment in it.
Somebody says a version of this sentence: “We can see the clicks. We just can’t see what happened after.” Then they look at me like the fix is a setting somewhere, or a report they haven’t found, or a feature we haven’t sold them.
It almost never is. In six years of standing this up, the break has been inside the advocacy platform maybe twice. Every other time it has been in the customer’s own plumbing, and the reason nobody found it is that advocacy attribution is not one measurement system. It is two separate data streams that only produce an answer when somebody deliberately joins them.
Most programs build one, assume it will eventually tell them everything, and then get frustrated when it doesn’t. Attribution isn’t unsolvable. It’s under-built.
Stream one is audience identity. When your employees post, real people at real companies choose to engage. Your program can know who those people are and where they work. That is a genuinely rare thing to have. It is the difference between “we reached professionals in financial services” and “these eleven people at this named account are paying attention to us right now.”
What this stream will never tell you is what any of them did next. It is a picture of attention, not of behavior downstream. Somebody senior at a target account reading your content is a real signal. It is not a form fill.
Stream two is click-through attribution. Tracking parameters on shared links, referral source in your analytics, source fields carried into your CRM and your ATS. This is the stream that knows about sessions, bounce, conversions, applications, opportunities.
What this stream will never tell you is who most of your audience was. It only sees people who clicked. And the majority of the value an advocacy program creates does not involve a click.
Think about what actually happens when a well-connected employee posts something good. Somebody reads it and doesn’t click. Somebody comments and gets into a conversation in the replies. Somebody DMs the employee directly. Somebody screenshots it into a Slack channel at their own company. Somebody remembers your name and Googles you nine days later, and arrives as organic search. Somebody’s boss brings you up in a vendor conversation because their team keeps seeing your people.
None of that leaves a referral trail. Dark social isn’t a measurement gap you can close with better tagging. It’s a description of how professional buying actually works, and the honest position is that a meaningful share of your program’s impact will always arrive unattributed.
So one stream sees the whole audience and none of the outcomes. The other sees a slice of the audience and all of the outcomes. Neither is the answer. The answer is the join.
These two streams are not two views of one database. They live in different systems, they are owned by different teams, and they are held together by two entirely different disciplines.
Stream one holds together on tagging discipline. Content has to be tagged consistently, by campaign, by audience, by business goal, from day one, by whoever curates it. Tagging is boring, it happens at the moment somebody is trying to get a post out the door, and it is the first thing that slips when a program gets busy. Untagged content is not lost data. It is data you can’t segment, which for reporting purposes is close enough to the same thing.
Stream two holds together on analytics ownership. Somebody at the customer has to build the filter, maintain the comparison segments, and make sure the source parameter survives every hop between the click and the record. That person almost never sits in the same org as the program owner. Frequently they have never heard of the advocacy program.
If nobody owns the join, there is no join. There is just two half-stories in two tools, presented to an executive who wanted one.
When a program tells me they see traffic but no outcomes, I stop looking at the advocacy platform and start walking the path a click actually takes. It is nearly always one of a small number of unglamorous failures.
The parameter drops on a redirect. The shared link points at a page that redirects, and the redirect is configured to hand off the path but not the query string. One global company I worked with had this on their careers site: every job link went through a regional redirect that quietly stripped everything after the question mark. Traffic arrived. It arrived as direct. Months of employee sharing landed in the analytics bucket labeled “we have no idea where this came from.”
The destination site strips parameters mid-journey. Consent management, a personalization layer, a login wall, or a site search step reloads the page clean. The first pageview is attributed correctly. Everything after it is orphaned, so the conversion at the end of the funnel is attributed to nothing.
The ATS or CRM never receives the source at all. A common ATS configuration treats the application form as its own hosted property, so the candidate’s referral source ends the moment they click “Apply.” The careers page knows advocacy sent that person. The applicant record does not. And the applicant record is the only one anyone reports on.
None of these are exotic. All of them are invisible unless somebody tests the actual path end to end, with a real link, on a real device, and confirms the value shows up in the system where the outcome gets recorded. That test takes about twenty minutes and I have never once regretted running it before launch.
Attribution is shared infrastructure, and infrastructure has named owners on both sides of the line.
On our side, the parameters have to be configured to match the customer’s own analytics taxonomy, not a generic default. On their side, somebody in digital analytics owns the website filter, somebody in marketing ops or RevOps owns whether source survives into lead and opportunity records, and somebody in talent ops owns whether it survives into the ATS. Four names, four systems.
Get those names in week one, or you will be trying to get them in week forty, from a program owner who now has to explain why they need them. We covered why measurement decisions have to be made at launch rather than at renewal in How to Measure Employee Advocacy ROI (Without Counting Shares); this post is the plumbing underneath that argument, not a second framework on top of it.
The good news is that this is a one-time configuration. Two weeks of coordination pays out every quarter for the life of the program. The bad news is that it is genuinely cross-functional, which means it needs an executive sponsor to unblock it. It will not happen because a program admin sent a polite email.
Once the plumbing is real, there is still one move left, and it’s the one that decides whether anyone funds you next year.
Absolute numbers do not persuade. “Advocacy drove 2,000 visits to the site” is a fine number and it means nothing on its own, because nobody in the room knows whether 2,000 is good.
Comparison persuades. “Advocacy traffic converts at a multiple of paid social, with a lower bounce rate and longer sessions, at no media cost” is a different sentence entirely. It’s the same underlying data. What changed is that you gave the CFO a benchmark they already fund.
So build the comparison segment at the same time you build the filter. Advocacy versus paid social, versus organic search, versus email. Baseline it against what your own channels actually delivered, because the moment your comparison rests on a number somebody in the room can dispute, the conversation becomes about the number.
Let me be straight about the ceiling, because overselling this is the fastest way to lose the room.
The most common over-promise in advocacy measurement is “this tells you who became revenue.” It does not. No attribution architecture does. You will not get a clean line from a specific person reading a specific post to a specific closed-won deal, and any vendor who implies otherwise is selling you a story you will have to defend later without them.
What you can build is a defensible floor. A known audience of named companies and roles choosing to pay attention. A measured, comparable set of outcomes from the traffic that did click. An honest acknowledgment that both of those undercount, and a program that stops pretending otherwise.
That floor is unglamorous. It’s also the thing finance actually funds, because it’s the only version of the number that survives being questioned.
Want the plumbing walked with you? Four names, four systems, and a twenty minute end to end test that has never once been a waste of time. Book a demo and we will walk your path before launch, not at renewal.
See the platform that comms, marketing, and compliance teams use to run their advocacy program.
