What is an Event Chain?
Attribution is the most argued-over number in marketing, and it earns the arguments, because in most firms it is not really a number at all. It is an assertion wearing one. A model decides in advance how credit will be shared among touches, the deals flow through, and the result is presented as a finding. Everyone in the budget meeting suspects this, which is why the meeting never ends.
There is a more honest way to hold the question. An Event Chain links marketing events through to revenue outcomes so that attribution is a chain of evidence, each link recorded and inspectable, rather than an assertion of credit.
A chain, not a model
Conventional attribution starts from a policy: first touch takes the credit, or last touch does, or a curve spreads it across everything in between. These are conventions, not evidence. They answer the question “how shall we share the credit?” and they will produce a confident answer whether or not anything connects the touch to the money.
A chain answers a different question: what actually connected this event to this revenue? An event happened; a person responded; the response became a meeting; the meeting became a proposal; the proposal became a commitment; the commitment became revenue. Each connection is recorded when it happens, so the chain can be walked in either direction: from the event forward to the money, or from the money back to what genuinely preceded it. It is the same instinct that decision provenance applies to decisions, applied to demand: keep the connective tissue, because the conclusion is only as good as the chain behind it.
Where chains break
Every real chain has weak points, and they cluster in predictable places. Handovers, first: marketing passes to sales, sales passes to delivery, and at each seam the connection survives only in someone’s memory or a free-text note. Offline moments, second: the relationship that actually advanced over a dinner, a referral, a corridor conversation at a conference no system observed. And time, third: the deal that goes quiet for a year and returns under a new name, with a new contact, at the same company.
The discipline is not to pretend the chain is complete. A broken link is recorded as broken. A bridge that is guessed is labelled for what it is: “the champion attended the event and the deal opened a month later” is inferred; “the client said they heard us on the webinar” is stated. In the language of the evidence state, a chain has the quality of its weakest link, and an honest record shows that quality rather than laundering every connection into equal certainty.
Honest attribution versus claimed attribution
Claimed attribution is complete, tidy and wrong in unknown ways. It is comfortable in a budget meeting precisely because the model fills every gap. Honest attribution is often shorter and always messier: some chains run clean from event to revenue, others fade into an inferred bridge, and some events end in silence. But honesty is what makes the record usable for decisions. Budget can be weighted toward the chains that hold under inspection. The gaps stop being invisible and become findable, fixable problems: record the handover, ask the source question at intake, close the seam between systems. And the firm stops rewarding whichever channel happens to sit closest to the model’s favourite position.
One concrete example
Clearly illustrative, with no customer implied. A firm runs an executive roundtable. Two chains follow. In the first, an attendee books a follow-up meeting the same day; the meeting becomes a proposal within a month; the proposal becomes a commitment. Every link is recorded, and most of the chain is measured. In the second, a deal opens four months later at another attendee’s company, through a different contact, and the only bridge between the event and the deal is inferred from the attendance record. An honest chain report shows one strong chain and one weak one, labelled as such. A last-touch model, meanwhile, would have handed the entire second deal to the website form the new contact happened to fill in, and the roundtable would have looked like a cost. Multiply that by a year of events and the model is not simplifying the truth. It is inverting it.
Chains end in outcomes, and outcomes teach
The end of an event chain is not the invoice. The commitment the chain produced gets delivered, and eventually scored, and that score lands in the Outcomes Ledger with the chain still attached. Over time the firm learns not just which events produced revenue, but which produced revenue that turned out to be worth winning, which is the version of attribution that actually deserves budget. That is the standard decision intelligence holds every claim to, and ONX holds marketing to the same one: events linked through to revenue outcomes, so attribution is a chain of evidence, not an assertion.
Common questions
What is an Event Chain?
An Event Chain is the linked sequence connecting a marketing event to a revenue outcome: event to response, response to meeting, meeting to proposal, proposal to commitment, commitment to revenue, with each connection recorded as it happened. It makes attribution a chain of evidence that can be inspected link by link, rather than an assertion produced by a credit-sharing model. Where a link is missing or inferred, an honest chain says so.
How is an event chain different from attribution modelling?
Attribution models are policies for sharing credit: first touch, last touch, or a curve across touches. They answer “how shall we distribute the credit?” by convention, whatever actually happened. An event chain answers a different question: what recorded connections actually run from this event to this revenue? A model produces a complete-looking number regardless of the evidence. A chain is only as long as the evidence, which is exactly what makes it useful.
Where do event chains usually break?
At handovers and offline moments. Marketing hands to sales and the connection survives only in memory or a free-text field. A relationship moves forward over a dinner, a referral or a corridor conversation no system saw. A deal dies and returns a year later under a new name. The discipline is not to pretend these gaps away but to record them: a bridge that is inferred is labelled inferred, and a claim the client stated is labelled stated.
What is honest attribution?
Attribution that shows its evidence and admits its gaps. An honest chain may be shorter than a modelled one, but every link in it holds under inspection, and the weak links are labelled with their evidence quality. That makes it more useful for decisions, because budget follows chains that survive scrutiny, and the recorded gaps become fixable problems rather than invisible ones.