Cross-promote without link schemes or spam

Independent founders should be able to help each other reach appropriate audiences. That does not make every exchange healthy. “Backlink for backlink,” forced follows, automated comments, undisclosed paid endorsements, and irrelevant widget networks create activity without creating reader value.

The clean boundary is intent plus implementation: would this recommendation still deserve to exist if search engines ignored the link? If the answer is yes because it helps a specific audience discover a relevant product, you have a useful starting point. If the answer is no, redesign it.

A fair credit can reward the work of hosting a recommendation. It must never purchase a promise of rankings, reviews, or artificial engagement.

Recommendation versus link scheme

A network needs explicit rules against manipulation because scale amplifies bad incentives. Project Relay’s intended unit of value is a reviewed placement, not a dofollow link, social follow, fake sale, or engagement event.

Disclose the relationship plainly

Readers should understand why a recommendation is present. Use language people recognize—“partner recommendation,” “sponsored placement,” or another accurate label—close to the creative. Do not bury the explanation on a terms page.

If money or another material benefit influenced the placement, publishers should apply the link attributes and disclosures appropriate to their jurisdiction and search provider guidance. A relay credit is a material exchange even when no cash moves between the two displayed products. Because requirements vary, founders remain responsible for the rules that apply to their site; the platform should provide conservative defaults and editable disclosure text.

Make relevance testable

“Both are SaaS” is not enough. A useful match connects a recognizable audience need to a supported capability without placing direct substitutes beside each other. A lightweight review can ask:

• What exact reader problem makes this destination useful here?
• Is the recommendation adjacent to relevant content or workflow?
• Could the pairing confuse readers about product ownership?
• Are the products direct competitors for the same primary job?
• Has the host explicitly approved this audience and category?

Context also determines format. A one-line sidebar card may suit a documentation page; a longer founder note may suit a resource collection. Networks should not require permanent homepage placement simply because it is easy to measure.

Measure without following people around

A placement ledger needs enough evidence to confirm that work happened. That does not justify building cross-site profiles. Aggregate impressions, aggregate clicks, timestamps, placement status, and coarse fraud signals are usually sufficient. Avoid third-party cookies, fingerprinting, session replay, or passing visitor identities between founders.

Publishers should be able to inspect the widget code and understand every network request. Visitors should be able to open the promoted product without first creating a Relay account. Account exports and deletion controls should cover founder data, while financial records can follow documented legal retention requirements.

Keep AI in the middle—not in charge

AI can profile public pages, suggest non-competing matches, draft bounded creative, detect unsupported claims, and flag suspicious placements. It should not auto-publish, fabricate personal testimony, negotiate away a founder’s constraints, or generate fake activity to settle a credit.

The safe workflow is evidence → draft → automated check → host approval → promoted-owner factual approval → immutable publication. Our companion guide explains how to build source-backed recommendations inside that workflow.

Know when not to participate

Decline a placement if the destination is misleading, unsafe, under construction, irrelevant, a direct substitute, or incompatible with your values. Also decline if the requested wording overstates your experience or if the disclosure requirement would undermine the requester’s goal. A well-designed relay treats rejection as quality control, not noncompliance.

Platforms should suspend participants who hide placements, automate clicks, repeatedly submit unsupported claims, impersonate owners, or pressure hosts. Appeals and audit history keep enforcement accountable.

The compact ethical playbook

1. Verify ownership, sources, and the live placement.
2. Match a real audience problem, not merely a category label.
3. Draft only from owner-approved evidence.
4. Disclose the partner relationship beside the recommendation.
5. Approve with a human on both sides.
6. Measure aggregate delivery without tracking identities.
7. Attribute links conservatively when value is exchanged.
8. Retire expired placements and preserve an audit trail.

This framework does not guarantee search performance, traffic, or sales. It creates something more defensible: a network where each founder completes a useful action before receiving one, with evidence that the action was real. See how the verified relay works end to end.