What we are not building

Partner program software is defined as much by what it refuses to do as by its feature list, and RelayWonder refuses five things on purpose: we will not hold or pay out partner money, we will not build an Amazon channel, we will not run your program for you as an agency, we will not charge a percentage of commissions, and we will not show fake activity anywhere in the product. Each of these is a request we have received, more than once, from people we would like as customers. This post explains the reasoning behind each refusal with the numbers that drove it, because a roadmap is only trustworthy when the things left off it are explained as carefully as the things on it.

Why partner program software needs a list of refusals

Every feature a small team ships has two costs. The first is building it. The second, larger and permanent, is that it now has to be kept working alongside everything else, and it shapes what the next feature can be. A product that accepts every reasonable request becomes a product that does nothing particularly well and is impossible to explain in one sentence. Our one sentence is: a partner network where brands track, pay and recruit on real data, and partners carry one account and one balance across every brand. The five refusals below are the ones that protect that sentence.

No managed payouts

The most frequent request is "just pay the partners for us". It sounds like a convenience feature. It is actually a change in what kind of company we are. To pay partners on your behalf we would have to collect money from you first, hold it, and then send it on. At that moment we are holding client funds, which in most jurisdictions makes us a payment business with licensing, capital and audit obligations, and makes you an unsecured creditor of ours for whatever balance is sitting with us. If we failed, your partners' money would be in our estate, not yours.

There is a second, quieter problem. A platform that holds money has an incentive to hold it longer. Float earns interest; delayed payouts smooth cash flow. We did not want to build a company whose economics improve when partners wait. So we compute, and you pay. A payout batch becomes a CSV in the exact format PayPal Payouts [1] or Wise batch payments [2] accept, you upload it to your own account, the money goes directly from you to the partner, and you mark the batch paid. The ledger shows the paid rows; we never see a cent of it. PartnerStack, by contrast, offers platform-managed payouts as part of its quote-based service as checked on 2026-09-26 [3], which is a legitimate choice for enterprise brands that want to outsource the whole function, and is not the one we made.

No Amazon channel

Brands that sell on Amazon ask whether their partners can earn commission on Amazon sales through us. The answer is no, and it will stay no. Amazon Associates is a closed program with its own operating agreement, its own tracking, its own cookie window, its own rate card by category and its own payment schedule [4]. An affiliate who wants to earn on Amazon sales joins Amazon Associates directly; a brand selling on Amazon has no way to see which affiliate drove which sale except through Amazon's own reports, and no way to pay a different rate than Amazon pays.

Everything RelayWonder does well depends on seeing the invoice: first-party cookie plus Stripe invoice-level attribution, refunds and disputes as negative rows, personal coupon codes that attribute with no click. None of that is possible when the checkout belongs to Amazon. We would be building a thin, unreliable layer over data we cannot access, and the first dispute between a brand and a partner over an Amazon sale would be one we could not adjudicate. A brand that sells on Amazon does not need us for that channel, and we would do it badly. That is two reasons, and either one is enough.

No agency services

"Can you just run it for us?" is the request we understand best, because recruiting is genuinely the hard half. Agencies that run affiliate programs typically charge a monthly retainer plus a share of commission, and for a brand with a large budget and no time that can be rational. We are not going to offer it, for a reason that is about the product rather than about agencies.

The product has to be usable by one person on a Tuesday afternoon. If we had a services team, every rough edge in the console would become a ticket for that team instead of a thing we fix in the product, because fixing it in the product helps all brands while the services team only helps the paying few. Within a year the product would be the thing our own staff work around. We have watched this happen to other tools in the category. So instead of an agency, we build the things an agency would do into partner program software that the brand runs: Hunter search for partners already in the network, content campaigns with a brief and a bounty, outreach drafts generated from your template that you copy into your own mailbox. The open-web search sources and AI drafting are not connected yet, and the screens say so in plain language; what we will not do is hide that gap behind a person on a retainer.

No percentage of your commission

Several platforms price as a percentage of the commissions that flow through them, sometimes alongside a monthly fee. It is appealing to the vendor because revenue grows automatically with the customer. It is a bad deal for the brand for one arithmetic reason: the brand's cost grows with the partner's success, and the vendor did nothing extra to earn it. Our pricing is a flat monthly tier by partner-driven revenue: free up to $1,000 per month, $49 up to $10,000, $129 up to $50,000, custom above that. Zero percent on every plan, and partners never pay anything. Here is what the difference looks like at five program sizes, assuming a 20% commission rate and a hypothetical percentage platform taking 5% of commissions plus a $49 base; the 5% is an illustration of the pricing shape, not a quote from any named vendor.

At $50,000 of partner-driven revenue a percentage model costs $549 per month in this illustration against $129 flat, and the gap widens from there. More importantly, a flat tier means the brand can raise commission rates to attract better partners without raising its software bill. We want brands to pay partners more, not to pay us more when they do.

No fake activity

This one has no revenue argument; it is simply a rule. Nothing in production shows a number we do not have. No placeholder notifications on first login, no demo partners in an empty list, no "estimated reach" badges computed from nothing, no "coming soon" buttons that open an empty modal. When a feature depends on a service we have not connected, the screen says what is pending and why in one sentence a person would use. InsightWonder does not estimate prompt volume, so the AI opportunities page shows no volume column at all rather than a guessed one.

The reason is that the whole product is a ledger of promises between brands and partners. If a brand learns that the dashboard contains one invented figure, every figure becomes suspect, including the ones that decide whether a partner gets paid. Trust in a commission ledger is binary. We would rather show an empty state with an honest sentence than a full-looking screen that teaches the user to doubt it.

What saying no makes possible

Each refusal buys something concrete. Because we do not hold money, the append-only ledger can be the single source of truth and the payout batch can be a plain file; there is no reconciliation between our balance and yours because we have no balance. Because we do not do Amazon, every attribution path ends at a Stripe invoice we can read, so refunds and disputes become negative rows automatically. Because we do not run an agency, every recruiting feature has to work as a button, which is why content campaigns and the MCP endpoint exist as self-serve tools. Because we do not take a percentage, the pricing page fits in one table and there is nothing to negotiate. And because we do not fake activity, a brand that sees a number can act on it.

• Ledger: append-only, invoice-level, with refunds and disputes as negative rows and a 30-day default hold.
• Payouts: monthly batches exported as PayPal or Wise CSV, paid from the brand's own account, marked paid by the brand.
• Recruiting: network search, campaigns with bounty and commission, outreach drafts for your own mailbox, an MCP endpoint with scoped keys.
• Pricing: four flat tiers, one of them free, no percentage, partners never pay.
• Honesty: every pending integration labelled; no estimated or placeholder numbers in production.

What is still on the roadmap

Saying no to five things does not mean the list is short. Everything below has to fit around the refusals, and each item is currently labelled as pending in the product rather than pretending to exist.

• Open-web Hunter search sources, so a brand with no partners can find candidates beyond the network.
• Mailbox connection, so outreach drafts send from your own account instead of being copied. Sending from our domain stays off the list.
• AI drafting of briefs and outreach from structured fields, and AI scoring of partner fit.
• Shopify checkout support alongside Stripe, with the same invoice-level rules.
• A partner-side MCP key so partners' own assistants can read their balances and compare programs.

If one of the five refusals is a deal-breaker for you, that is useful to know early, and we would rather point you to a vendor that made the other choice. PartnerStack manages payouts; Partnero, Rewardful and FirstPromoter each have strengths we list on our comparison pages with the dates we checked them. Partner program software is a choice about what you want the tool to be responsible for, and we have tried to make ours easy to read.