Most affiliate pipelines are a list of names in a tool that supports stages. That is not the same as a pipeline, and the difference shows up in revenue.
A list tells you who exists. A pipeline tells you what to do this morning.
Stages worth having
Use stages named after things that actually happen in your process. If a stage does not change what you do next, delete it.
For most affiliate businesses, this is enough:
New. They arrived, nobody has spoken to them. Contacted. A real first message went out. In conversation. They replied. This is the only stage where the deal is genuinely alive by someone else's choice. Decision. They have what they need and are deciding. Closed, won or lost.
Five stages. Anything more elaborate is usually imported from an enterprise sales process and will quietly stop being updated within a month, at which point the pipeline is worse than useless because it looks authoritative and is wrong.
The two numbers
Forget conversion rates for a moment. Two counts predict revenue better than anything else, and both are embarrassingly simple.
How many leads have had no contact at all. Every one is a lead you paid for and did not use. The response-time research is clear that value decays fast, so this number is not a backlog, it is a write-off in progress.
How many have not moved in fourteen days. These are the ones that quietly die. Nobody decides to abandon them; they just stop being the most urgent thing, every day, until they are cold.
If those two numbers are near zero, the pipeline is working, almost regardless of what the conversion rate says. If they are not, no amount of new traffic helps, because the new leads will join the same queue.
Why pipelines rot
Updating is manual and boring. If moving a lead between stages takes several clicks in a tool you have to remember to open, it will not happen on a busy day, and busy days are the ones that matter.
The stages do not match reality. Someone books a call and there is no stage for that, so they sit in "contacted" for a fortnight looking identical to someone who ignored you.
Nothing is ever lost. Marking a lead lost feels like admitting failure, so the pipeline accumulates. Six months in, half the entries are dead and the view no longer means anything.
The fix for the first is structural: the pipeline should update itself where it can. A booking should move the lead. A reply should move the lead. If your system knows those things happened and does not act on them, you are doing bookkeeping the software could do.
Working it, daily
The pipeline earns its keep in one habit: a short pass, at the same time each day, answering one question. Who needs something from me before this evening?
That is usually three to six people. New leads with no contact, replies waiting on you, and anything at the decision stage that has gone quiet. It takes ten minutes if the pipeline is honest, and it is the highest-value ten minutes in the day.
Everything else, including the reporting, can wait for a weekly pass.
What changes when you grow
The first hire, or the first partner, breaks a pipeline that lived partly in your head. Suddenly the unwritten context matters: who has already spoken to this person, what was promised, whether this lead came from the expensive channel or the free one.
Two things need to be true before that transition, and both are easier to establish now than later. Every lead needs an owner, and every lead needs its source attached from capture. Without the first, leads fall between people. This is exactly what shared templates, scripts and a single pipeline view are for, and it is worth building the habit while the pipeline is still small enough to fix cheaply.
Reading a pipeline in thirty seconds
A pipeline that takes analysis to interpret will not get looked at. The test of a good one is whether the shape tells you something before you read any names.
Top-heavy. Lots in new, few further along. Either leads are arriving faster than you can contact them, or the first message is not landing. Check the untouched count: if it is high, it is capacity; if it is low, it is the message.
Middle-heavy. Everyone is in conversation and nothing closes. Usually the ask is missing. People rarely stall because they need more information; they stall because nobody asked them to decide.
Empty at the top, busy at the bottom. Comfortable and dangerous. You are working the current cohort well and generating nothing behind it. This is the shape that precedes a bad quarter, and it is invisible if you only look at revenue, because revenue is still fine when it appears.
Flat. Roughly even across stages, moving steadily. This is what a working pipeline looks like and it is rarer than you would think.
Stage-change hygiene
Two rules keep a pipeline honest, and both are unpopular.
Every lead has a next action with a date. Not a status, an action. "Follow up Thursday" is a next action. "In conversation" is not. A lead without one is either finished or forgotten, and marking which is the whole job.
Nothing moves backwards silently. If a lead in decision goes quiet for three weeks, that is information, and dragging them back to "in conversation" without noting why destroys it. Note the reason. Patterns in those reasons are the most useful data the pipeline produces, and almost nobody captures them.
The weekly pass
The daily pass answers who needs something today. A separate weekly pass, fifteen minutes, answers different questions.
Which leads have not moved in a fortnight, and are they genuinely alive? Which source produced the leads that reached decision this month, as opposed to which produced the most leads? What is the most common reason people stalled?
That third question is the one that changes the business. If four people this month stalled on the same objection, that objection belongs in the follow-up sequence, on the landing page, and in the first conversation. Most pipelines contain that answer and nobody goes looking for it.
Owners, before you need them
Assign an owner to every lead now, while the answer is always you.
It sounds like pointless bureaucracy for a business of one, and it is, right up until the week you bring in help and discover that "who has already spoken to this person" is knowledge that exists only in your memory. At that point retrofitting ownership across four hundred leads is an afternoon of guessing.
The same argument applies to source tagging, and to writing down what you promised rather than remembering it. None of it pays off today. All of it is enormously cheaper to establish now than at the moment it becomes urgent, which is invariably a moment when you are already busy.
What good looks like
A pipeline is working when three things are true at once. Nobody has been sitting untouched. Nothing has been stationary for a fortnight without a reason recorded against it. And you can answer, without opening a report, which source is producing the leads that actually close.
Most pipelines fail all three, and the fix is rarely a better tool. It is a smaller pipeline, honestly maintained, worked for ten minutes a day.
Common questions
What stages should an affiliate pipeline have?
Four or five, named after things that actually happen: new, contacted, in conversation, decision, closed. If a stage does not change what you do next, it is a label rather than a stage.
How do I know if my pipeline is healthy?
Two numbers. How many leads have had no contact at all, and how many have not moved in fourteen days. Both should be near zero, and in most pipelines neither is.
How many leads should be in an affiliate pipeline?
Fewer than most people keep. A pipeline full of leads you will never contact is not optimism, it is noise that hides the ones that are live.
When should I mark a lead as lost?
When you have stopped intending to contact them. Leaving them open costs nothing in the tool and costs a lot in clarity, because it makes the pipeline look busier than the business is.