When to stop paying attention to it

By Jimi Barkway · Published 1 September 2026 · Part of the honest manual

Here's the module a software vendor isn't supposed to write. At some point, probably within the first year, your program will settle: a handful of producing partners, a steady trickle of referred revenue, not much movement month to month. Every piece of content in this category tells you that's the moment to push harder, run enablement campaigns, hire an affiliate manager.

Usually it isn't. Usually the right move is to put the program on autopilot and go build your product, and this module is permission to do that, plus the checklist that makes it safe.

A plateau is an achievement wearing a disguise

Run the numbers on a "stalled" program: $1,500 a month of referred revenue at a 20% commission costs you $300 and roughly none of your time once it's automated. That's a customer-acquisition channel with a fixed cost of nearly zero that never sleeps and never asks for a raise. If a consultant offered you that, you'd take it. The only reason it feels like failure is that you once imagined the graph going up forever, and graphs don't.

The founders who get burned here are the ones who treat the plateau as a crisis: they mass-recruit partners who'll never produce (remember the median: around 15% ever refer anyone), churn out assets nobody asked for, and burn a quarter's attention on their fourth-best channel. The program didn't need saving. It needed leaving alone.

What autopilot actually means

Autopilot isn't abandonment. It's four specific things, two of them run by machines:

• Payday runs itself. Automated payouts on the schedule you promised, because module 3's rule doesn't relax when you're busy: a missed payday is the one failure partners never forgive.
• Onboarding runs itself. The welcome-and-first-action sequence from module 5 keeps firing for every new partner who wanders in.
• You reply to partners within a day or two. Ten minutes a week. A partner who asks a question and hears silence tells the story publicly.
• The ten-minute monthly check, which is the whole of the next section.

The ten-minute monthly check

The three numbers from module 6, once a month, in a note you keep:

• Producing partners. How many affiliates generated anything. Watch for one going quiet who usually doesn't; one personal message to them is the highest-value retention work that exists.
• Kept referred revenue, net of refunds. The trend over three months matters; any single month doesn't.
• 90-day retention of referred customers against everyone else. If it ever drops well below organic, something upstream changed and it's worth ten more minutes.

Glance at any risk flags while you're in there, note the three numbers, close the tab. Done properly this is genuinely ten minutes, and months of these notes become the dataset that answers the next section's question.

Plateau or ceiling? The difference is in the data

Sometimes the settle-point really is money left on the table, and your monthly notes will say so. Reinvest when you see a repeatable pattern:

• One kind of partner clearly works, and more of that kind exist. Your producing partners are all newsletter writers in your niche? There are fifty more newsletters like theirs, and module 4's playbook runs again on a warm-er list.
• Inbound applications are rising on their own. Something out there is working; find out what and feed it.
• Referred customers retain better than organic. The channel is bringing your best customers, which is the strongest argument that exists for giving it more attention.

And it's probably a real ceiling when growth only ever happens in months where you personally pushed, when new partners produce at or below the median rate no matter who they are, and when your "top 10%" is one person. Pushing against that ceiling costs the same energy as building product features, and the features compound.

That's the manual

Nine modules, no gate, no email required, and if you read them in order you now know more about running a small affiliate program than most of the people paid to write about it. Two standing invitations. If a figure anywhere in the manual has gone stale, tell us at support@affiliaterail.com; every number here carries its checked-on date for exactly that reason. And if the module you needed doesn't exist, say so, because the next one gets written where the questions are.

Now close the tab and go build the thing your affiliates are going to sell.