Content campaigns: paying partners for the page, not just the sale

A content campaign is a way to pay a partner for a specific page or video, not only for the sales it later produces. It has three parts: a brief that names the buyer question to answer, a bounty paid once per delivered piece, and ordinary affiliate commission on anything the piece sells afterwards. Brands run campaigns because commission alone produces whatever partners feel like making, and what partners feel like making is rarely the one page that answers "which tool for X" with your name in the first paragraph. That page is the one an AI assistant cites, and nobody writes it by accident.

This post lays out what goes into a campaign, what a fair bounty looks like with the arithmetic shown, how the run loop works inside RelayWonder, how we check whether the delivered page is being cited, and where campaigns fit next to plain commission and old-fashioned sponsorship.

Why commission alone does not get you the page

Picture a partner with a newsletter of 8,000 subscribers and a 20% recurring commission from your program. The rational move for them is to mention you in a roundup of six tools, because a roundup can carry six affiliate links and six commissions. The irrational move, from their point of view, is to spend a day writing a 1,500-word piece that answers one narrow question about your category, links only to you, and will earn commission from perhaps three sales a month.

From your point of view the narrow piece is worth far more. It is the page that matches a real buyer question word for word, and it is the page that an assistant will pull into its answer when someone asks that question. Our March 2026 sample of 200 buyer questions across SaaS categories found that the cited sources were overwhelmingly single-topic pages, not roundups; we say "our sample" because it was our sample, and you should run your own on your own category.

The gap is a payment problem, not a motivation problem. The partner is doing the sensible thing given how they are paid. A bounty per piece changes what the sensible thing is.

What a content campaign contains

In RelayWonder a campaign is a form with seven fields. All of them are visible to partners before they claim a slot, so nobody discovers the deadline after they have started.

• Topic: a real buyer question, written as the buyer would type it. "Best way to track affiliate renewals in Stripe" is a topic. "Awareness content for Q3" is not.
• Why this topic: one or two sentences on what the brand is losing, for example that assistants currently cite a competitor or cite nobody. Partners write better when they know the stake.
• Accepted formats: any of YouTube, TikTok, blog, newsletter, Reddit. A brand can accept one or several; the bounty can differ per format.
• Bounty per delivered piece: a fixed amount paid when the brand confirms the delivered URL. It posts to the partner's ledger like any other row and follows the same hold and minimum.
• Commission on sales: the partner's normal rate, tracked through their normal link and personal code embedded in the piece.
• Cap: the maximum number of pieces the brand will pay bounties for. Three to five is typical for a narrow question; one is fine for a single definitive video.
• Deadline: a date after which unclaimed slots close and claimed but undelivered slots are released.

There is deliberately no field for "brand messaging" or "required talking points". A brief that dictates sentences produces a page that reads like an advert, and advert-shaped pages are exactly what assistants and readers skip. The brief states the question and the facts; the partner decides the words.

Pricing the bounty with real arithmetic

The bounty should cover the partner's production cost for a piece that only makes sense for one brand, and no more, because commission handles the upside. Here is the comparison a brand can run for a single buyer question where it wants three blog posts and one YouTube video.

Run the numbers forward. Suppose the four pieces together bring eight customers in the first quarter on a $79/month plan, and six of them are still subscribed after a year. Attributed revenue over twelve months is roughly 8 × 79 × 3 for the first three months plus 6 × 79 × 9 for the rest, which is $1,896 plus $4,266, or $6,162. Commission at 20% is $1,232. Total paid to partners: $750 in bounties plus $1,232 in commission, $1,982, against $6,162 in revenue, a 32% effective cost that falls every month the customers stay. The sponsorship route would have cost $1,500 or more for a single video with no commission motive after launch and no second, third or fourth piece.

The numbers go the other way too. If the four pieces bring nobody, the brand is out $750 and has four pages that answer a question it cares about, which is still not nothing. The bounty is sized so that a miss is affordable.

Choosing the topic: cite competitors, cite nobody, cite you

The best campaign topics are buyer questions where an AI assistant currently gives a bad answer for you. There are three states for any question, and each suggests a different campaign.

1. The assistant cites a competitor. This is the most urgent gap and the one where a single-topic page from a credible partner has the clearest effect. Topic: the exact question, with a brief that includes your honest comparison points.
2. The assistant cites nobody, or answers vaguely. Here there is no incumbent and a well-structured page can become the reference. Topic: the question plus a request for a worked example or a checklist, because those are the structures assistants quote.
3. The assistant already cites you. No campaign needed for that question; spend the bounty elsewhere. Re-check quarterly, because citations move.

Brands that use InsightWonder, our sister product, get these three states for their category delivered into RelayWonder on an hourly sync, and can open a campaign from a gap with the question pre-filled. InsightWonder does not estimate how many people ask each question; it reports who is cited. Brands without it can do the same by hand: ask three assistants twenty questions and record the sources. It takes an afternoon.

How a campaign runs from publish to paid

Once the brand publishes, the loop is short and every step leaves a row.

1. Eligible partners see the campaign. Eligibility is by partner type and declared formats: a newsletter writer does not see a TikTok-only brief.
2. A partner claims a slot. The slot is held for the deadline; nobody else can claim it, and the partner cannot claim more than one slot per campaign unless the brand allows it.
3. The partner publishes and submits the URL. The submission form asks for the URL, the format, and a confirmation that the piece carries a disclosure and the partner's tracking link or code.
4. The brand reviews. The review has three outcomes: confirm, request a change with a note, or decline with a reason. A declined slot is released. Most reviews we see take under a day because the brief was specific.
5. On confirm, the bounty posts to the partner's ledger as a row with the campaign id and the URL. It follows the same 30-day hold and $50 minimum as commission, and appears in the next monthly payout batch the brand exports.
6. From that moment the URL is tracked: clicks through the partner's link, sales attributed to it at the invoice, and whether assistants cite it.

The brand never pays the bounty through us. RelayWonder does not hold money. The row sits in the ledger, the brand pays from its own PayPal or Wise account using the exported batch, and marks it paid.

Review criteria that keep campaigns honest

A review should be a checklist, not a taste test, so partners know what "done" means before they start. We suggest brands use five checks and publish them in the brief.

• The page or video addresses the question in the brief in its title or first thirty seconds.
• It contains a disclosure of the paid relationship that a reader would actually see. The FTC endorsement guides are the reference for US audiences [1].
• Affiliate links are marked rel="sponsored" where the format allows, in line with Google's guidance on qualifying outbound links [2].
• The claims about the product are accurate. A partner who says the brand supports something it does not is corrected, not paid.
• The partner's own link or code is present so sales attribute.

Nothing in the list is about tone, length or whether the partner was enthusiastic enough. Those are the partner's craft, and the commission gives them a reason to be good at it.

Checking for AI citations after delivery

The reason a brand ran the campaign was to change what an assistant says. So after delivery we check. For each delivered URL, RelayWonder records whether the page has appeared as a cited source for the campaign's question in the assistants we monitor, and when it first appeared. When the gap that started the campaign begins to show the delivered page as a source, the gap is marked resolved on the brand's dashboard, with the campaign and the partner credited.

Two honest limits. First, a citation check is an observation, not a guarantee; assistants change their sources and we report what we saw on the date we saw it. Second, newer pages are cited less often than established ones in the first weeks, and the lag we see on our network is anywhere from a few days to a couple of months, with video transcripts and Reddit threads often appearing sooner than fresh blog posts. Perplexity and OpenAI both publish help material on how their products use and show sources, which is worth reading before setting expectations [3][4].

Formats and what each one is good for

The five formats behave differently in search, in assistants and in sales, and a brief should pick the ones that fit the question rather than asking for everything.

What is not in the product yet

We do not generate briefs or draft content with AI, and we do not score submissions automatically. A human at the brand writes the brief and a human reviews the piece. We may add drafting help later; for now, we think the brands who write their own briefs get better pages, and a scoring model would push partners toward whatever the model likes rather than what buyers ask. We also do not send emails to partners on your behalf about campaigns; the campaign is visible in their account, and if you want to nudge a specific partner, you copy a draft into your own mailbox.