Prioritisation
How to prioritize sales leads without a points system
Short answer
Every framework for this question is a set of initials. BANT, CHAMP, MEDDIC, and a dozen more. They all do the same thing: give you criteria, tell you to assess each lead against them, and produce a number.
Then you sort by the number and work down the list.
That process has one property nobody mentions, and it is the reason your numbers have not improved: a points total makes no claim that reality can contradict.
Why point systems feel rigorous and are not
Say a lead comes out at 87 and does not convert. What did you learn?
Nothing. Nobody claimed 87 meant they would convert. It meant they were more likely than a 64. When they go quiet, the system is not wrong, because it never said anything specific enough to be wrong.
Run that forward a year. You have contacted hundreds of people, assigned every one a number, and you cannot point to a single case where the model was disconfirmed. That is not a track record. It is a ritual that produces an ordering.
Compare it to a different kind of claim: this person said last Tuesday that their team cannot keep up with renewals.
That one can be wrong in several ways. They might have meant something else. It might have been someone else's problem. It might have been solved by Friday. Every one of those is checkable, and every check teaches you something about your own judgement.
The claim that can be wrong is the one worth making. It is the same logic as the falsifiability test in why your outbound stopped working, applied to the order of your list rather than the wording of your message.
The three tiers
Forget criteria. There is one question, and it sorts everything:
How specific is my reason for contacting this person, and where did it come from?
That gives you three tiers. They are not equal in size and they are not meant to be.
Tier 1 · A named reason
Something this specific person did or said, that you can point at. They wrote about the problem. They asked a question about it in public. They told you directly. You could paste the evidence into a message and it would make sense to them.
Tier 2 · An inferred reason
Something about their situation makes the problem very likely, but they have not said so. A new person in a role that owns this. A company that just changed something upstream of it. The inference is reasonable and specific to them, but it is still your inference, not their statement.
Tier 3 · A category match
They look like your other customers. Right size, right market, right job title. That is all. Nothing about this individual is doing any work in the decision.
Work Tier 1 completely before touching Tier 2. Work Tier 2 completely before touching Tier 3. In practice most people never reach Tier 3, and that is the correct outcome, not a failure to get through the list.
A worked example on 50 names
Here is what the split usually looks like. The numbers below are illustrative of the shape, not a result from any one account.
| Tier | Typical share of 50 | What you actually have | What you can open with |
|---|---|---|---|
| 1 · Named reason | 3 to 6 names | A dated thing they said or did | The thing itself |
| 2 · Inferred reason | 8 to 15 names | A situation that implies the problem | The situation, stated as a question |
| 3 · Category only | the rest | A profile match | Nothing specific, which is the point |
The first reaction to that table is that it feels wasteful. Forty names in Tier 3, and you are not contacting them.
But look at what Tier 3 actually is. If the only true sentence you can write is that they resemble people who bought from you, then every message you send them is interchangeable. You are not prioritising forty leads. You are declining to send forty messages that would have been identical.
Tier 1 · Named reason
3 to 6
Their words, dated, checkable
Tier 2 · Inferred reason
8 to 15
Specific to them, but still your inference
Tier 3 · Category only
the rest
True of hundreds of companies
188 of 6,400+
items our own pipeline kept after checking, a keep rate under 3 percent
Trendle internal pipeline totals, 17 July 2026
That is the same ratio from the other direction. When we check plausible-looking reasons at volume, under three percent survive. A hand-built list will do better than that because a human pre-filtered it, but not by as much as you would hope.
What to do when you have no conversion history
This is the case most advice ignores, and it is the case most people are actually in. You cannot weight criteria by past performance if you have no past.
The good news is that reason-based ordering never needed history. Judging whether you have a named reason or a category match is a judgement about the evidence in front of you, available on day one and on day one thousand.
What history eventually buys you is a check on your own standard: after fifty contacts you can look back and see whether the things you called Tier 1 behaved differently from the things you called Tier 2. If they did not, your Tier 1 bar is too loose. That is a real finding, and no points total will ever hand it to you.
The middle tier is where the judgement lives
Tier 1 is easy. Tier 3 is easy. Tier 2 is where people go wrong, in both directions.
Too generous, and Tier 2 swells until it is Tier 3 with a story attached. The test: could this inference be true of a hundred other companies? If yes, it is a category, not an inference.
Too strict, and you throw away the tier that carries most of the recoverable volume. An inference does not need to be certain. It needs to be specific to them and stated honestly. Opening with a question rather than an assertion is usually the right move, because it is accurate about what you actually know.
What Tier 1 evidence actually looks like
The tiers only work if your bar for Tier 1 is strict. Here is where the line sits, using the kind of thing people actually find.
| What you found | Tier | Why |
|---|---|---|
| They wrote a post describing the problem, three weeks ago | 1 | Their words, dated, checkable |
| They asked a public question about how to solve it | 1 | Stronger than a statement: they are already looking |
| They replied to someone else's post about it, in detail | 1 | Detail is the tell. A one-word reply is not evidence |
| A new person took over the function that owns this | 2 | Reasonable inference, specific to them, not stated |
| Their company changed something upstream of the problem | 2 | Same. The inference is yours, so ask rather than assert |
| They are a 50-person company in your best-performing market | 3 | True of hundreds of companies |
| Their profile lists a title that usually owns this | 3 | A title is a category with a person attached |
The two rows people argue about are the one-word reply and the job title.
A short reaction to a post is not evidence, because you cannot tell what they were agreeing with. Detail is what separates a person who has the problem from a person who recognises the topic.
And a job title is the most seductive Tier 3 signal there is, because it feels individual. It is not. It tells you what someone is responsible for, never whether the thing is currently going wrong.
The weekly ritual that keeps the tiers honest
Tiers decay. Left alone, everything drifts upward, because Tier 1 is where the interesting work is and nobody enjoys writing Tier 3 next to a name they like.
Fifteen minutes a week fixes it.
Pull the names you called Tier 1 in the past week. For each, read your stated reason without looking at the person. Ask the swap question: would this sentence be true of someone else on the list?
Every one that would be was misfiled. Move it down and note why. After a month you will have a short list of the specific ways you fool yourself, and those patterns are more valuable than the reclassification, because they are what you correct at the moment of judgement rather than a week later.
The number to watch is what share of Tier 1 survives the weekly review. If it is climbing, your judgement is improving. If it sits below half, you are not applying a standard, you are labelling your enthusiasm.
Where this does not apply
Where reason-based ordering is the wrong tool.
If you sell something cheap to a very large market, where the cost of a send is near zero and the decision is near-instant, sorting carefully costs more than it saves. Volume genuinely wins there. This page is about the case where a contact is expensive, either in money or in the goodwill you burn.
If you already have hundreds of closed deals and a functioning model built on them, use it. A model trained on your own outcomes beats a heuristic. This is a method for people who do not have that yet, which is most people, for longer than they expect.
And if your list is small enough to work end to end in a week, prioritisation is a distraction. Contact all of them and spend the saved effort on the messages.
What this looks like once it is running
Your list stops being a queue and starts being three queues with different rules.
Tier 1 gets your time and a specific opening. Tier 2 gets a shorter message that asks rather than asserts. Tier 3 gets nothing until something changes, and you keep it precisely so you notice when something does.
The number worth watching is not conversion rate. It is what share of your list makes Tier 1 over time. If that share is rising, your sourcing is improving. If it is flat while your list grows, you are adding volume without adding reasons, which is the most common way outbound gets worse while looking busier.
Related reading: how to say why you are reaching out covers turning a Tier 1 reason into an opening line, and should you buy a lead list covers what happens when the whole list arrives as Tier 3.
Questions people ask next
- How do I prioritize leads with no historical conversion data?
- Rank by reason specificity rather than by predicted probability. You cannot estimate conversion rates without history, but you can always tell whether you have a named, checkable reason for contacting someone or only a profile match. That judgement needs no data and improves as you make it.
- Is lead prioritisation by points a bad idea?
- It is not wrong so much as untestable. A points total never makes a claim you can check, so it cannot be disproven and therefore cannot improve. Reason-based ordering makes a specific claim about each person that reality can contradict, which is what lets you learn.
- How many leads should be in my top tier?
- Far fewer than feels comfortable. On most lists the tier with a genuine named reason is under ten percent. If a third of your list qualifies as top tier, your standard for what counts as a reason is too loose.
- Should I ignore the bottom of my list entirely?
- Not permanently. Names with no reason today are not disqualified, they are unripe. Keep them and re-check when something about their situation changes. What you should not do is contact them now just because the list looks thin at the top.