Ads Research Lab
A Project Moosed research lab for ad design and the future of ads and marketing.

The lab where advertising gets rebuilt for humans and agents

This is a standing research lab, not a marketing page and not an ad platform. The lab is where Project Moosed researches, designs, prototypes, measures, and publishes a better system of advertising, in the open, one cycle at a time.

Primary goal

Design and implement effective ads for humans and agents.

Same business, same campaign, two lanes. Human ads carry creative, brand, and feeling to human eyes. Agent ads carry structured, specific, takeable value to the software that increasingly shops on a person's behalf. The lab works on both lanes, and on the standards that let the agent lane run with permission, disclosure, and a log. Read The Model for the working definitions.

The method: a constant loop

Research→ Design→ Implement→ Measure→ Publish→ Research again

Every cycle should ship a better system than the last. Findings are public, graded, and dated. Claims on this site carry the Moosed grading practice (Confirmed, Reported, Contested) where they rest on outside evidence, and experiment pages publish only what was actually run and logged.

Current cycle status

Lab charter
Filed 2026-11-16
Paper #1
Agents and Advertising - in review (see Papers)
First experiment
Not yet carded. It will be carded from Paper #1 and its engineering requirements.
Latest publication
None yet. Paper #1 is in review.

Why a lab, and why now

Personal agents and AI browsers increasingly stand between an advertiser and a person. An agent does not consume advertising the way a human eye does: it summarises, extracts, blocks, or ignores creative whose only payload is persuasion. The lab's working thesis, stated as a theory to be tested rather than a result, is that advertising splits into a human lane and an agent lane, and that the agent lane runs on takeable value, verified identity, and logged redemption. Paper #1 reviews what others have said on this topic, marks where the evidence is strong and where it is thin, and states the model as numbered, testable propositions.

The lab is also working on the publisher side of the same problem: agent traffic costs a site real resources while the old benefits of a visit (an impression, a click, attribution) may not occur. The lab looks for a helpful solution to that, not a hostile one: dual-lane serving and standards-based agent access, so an agent visit can carry measurable value back to the publisher. That work will be published here as it is tested.

Relationship to Moosed Core

The Moosed Core ads platform is the production serving and billing system for local marketing. This lab is its research and development arm: what the lab proves, Core can adopt. Lab prototypes run on lab and test surfaces only. Nothing on this site switches on any ad serving, and this site carries no advertising itself.

For agents reading this site

This site serves agents properly from day one, because a lab about agent ads should eat its own cooking. The open agent layer is documented on the Standards page and includes a site manifest, a publications feed, and a plain text summary at /llms.txt.