Written by Brook Schaaf

This has been the warmest winter I recall here in Texas. Today, it feels like spring; I wore shorts when I took a walk with my wife and toddler, wiggling in his backpack. The unseasonable weather made me reflect on what happens when what we’ve grown accustomed to is different from what we expected. Attribution has been much on my mind as an example of this. I believe that all attribution is ready for an overhaul in the age of LLMs, representing both a meaningful opportunity and a serious threat to affiliate marketing. 

Why an overhaul? LLM mentions have broken—and are continuing to break—the heretofore recognized compensation methods that have been with us since multi-touch attribution was made possible by the internet. In order of appearance, as best I can tell, from the late 90s to early 00s: 

• pay per mille (thousand impressions) 

• pay per action

• last to arrive but strongest by far, pay-per-click 

Public Relations, sometimes unfairly reported as dead like its distant, now oft-kissed cousin, affiliate marketing, never went away—but its play-to-play nature has, thankfully, come to the fore.

In the journey across time and screens, through various gates like privacy blockers, the humble affiliate-associated transaction often fails to cross the tracking threshold. Some reward sites report tracking failures of half or more. Tracking platform data seems to confirm this, hence Awin’s CPI, as do occasional case studies from companies like Moonpull. This long-standing problem is compounded by LLMs, which effectively rob affiliate-supported content of any compensation. This content is necessary for inference, so “LLM takes everything, offers nothing” is neither fair nor sustainable. 

If we take, then, that this content serves as a sort-of impression and that there is value to each impression, even if no click or conversion can be associated therewith, then perhaps online advertising can roll up and fan out, as the LLMs themselves do. The roll up would be a single KPI, and the fan out would be contributing or associated factors, such as tracked transactions. 

For example, instead of cost per mille (thousand impressions), or CPM, we might consider Cost per Productive Mille, or CPPM. While this particular example is perhaps unorthodox, what’s important is the concept: there probably needs to be a new, single, compressed KPI. The additional letter proposed here captures the extra-dimensionality of the content value, which effectively teleports across time and screens past gateways to the user’s eyeballs.

Partnerize and impact.com have already taken meaningful steps in this direction, which others are likely to follow. A CPPM, with attached productivity measures (estimated or determined mentions, clicks, or transactions), could be provided not only by the tracking platform but also by the publisher and perhaps the merchant themselves. This would provide a point of triangulation as a basis for negotiation of compensation, which would ultimately be provided by the advertiser. This is the opportunity. 

The threat is that no such metric is created or honored, in which case content creators would continue to be starved, turning all of 2026 and beyond into winter.

The post CPPM appeared first on FMTC.