Routing leads without a round-robin

Round-robin is fair in the way that assigning surgeries alphabetically is fair. Everybody gets the same number of them and nobody is accused of favoritism, which are both real organizational goods, and it is completely indifferent to whether the outcome is any good.

The specific damage is subtle. Round-robin does not distribute leads randomly across reps, it distributes them randomly across rep-lead fit, which means your rep who has closed eleven staffing deals gets a manufacturing lead one time in four. That lead converts worse than it would have with the rep who knows manufacturing, and the rep learns slightly less than they would have from another staffing conversation. The loss compounds in both directions and appears nowhere in any report.

What to route on, in order

You do not need a machine-learning model. You need three inputs and a tiebreak, and the ordering matters more than the sophistication.

Segment fit comes first and is worth more than everything else combined. If a rep has closed multiple deals in an industry, route that industry to them. This is not a reward, it is a conversion decision: they know the vocabulary, they know what the objections will be, and they have references the prospect will recognize. In most teams this single rule captures the majority of the available gain.

Territory or relationship history is second. If anybody at your company has ever spoken to this account, the lead goes to that person, full stop, even if they are not the obvious owner. A prospect who calls back a company and gets a stranger who does not know the history experiences that as an organization that does not talk to itself, which is exactly the thing they are worried about when buying software.

Current capacity is third and is a constraint rather than a preference. A rep with forty open opportunities should not receive a forty-first regardless of fit, because the marginal lead will get worse attention than it would somewhere else. Capacity should throttle, not select.

Then a round-robin tiebreak among whoever remains, which is where round-robin belongs: as the last step among genuinely equivalent options, not as the first step among wildly unequal ones.

The part everyone gets wrong

Speed beats fit. If your routing logic takes eleven minutes to decide, your fit-optimized assignment is worth less than a dumb assignment made in ten seconds.

Inbound lead response curves are brutal and well documented. The difference in connect rate between a response inside five minutes and a response inside an hour is not a percentage point, it is a multiple. Any routing system that introduces latency to make a better decision is trading a large certain loss for a small probable gain.

So the design constraint is that routing must be synchronous and must never block on an enrichment call to an external vendor. Score on what you already know, assign immediately, and let enrichment arrive afterwards and reassign only if it changes the answer materially. Most teams build this the other way around, waiting for enrichment before assigning, and then wonder why their inbound converts poorly despite good rules.

The second thing everyone gets wrong is having no fallback. Every routing system eventually receives a lead that matches nothing: unknown industry, no territory, all reps at capacity. If the rule set does not terminate, the lead lands in a queue nobody owns, and a lead in an unowned queue is functionally a lost lead. Every router needs a default human at the end of the chain, and somebody should be looking at how often the default fires, because a rising default rate is the earliest signal that your segmentation no longer matches your market.

Instrument the assignment, not just the outcome

Here is the piece that makes the whole thing improvable rather than merely opinionated.

Log why each lead went where it went. Not just the assignment, the reason: matched on segment, matched on prior relationship, capacity override, fell through to default. Four values in a field, written at assignment time.

With that field you can answer the only question that matters, which is whether your routing rules are actually producing better outcomes than random. Compare conversion for segment-matched assignments against default-routed ones. If they are the same, your segment rules are decoration and you should stop maintaining them. If segment-matched converts meaningfully better, you have a number to defend the system with the next time somebody proposes going back to round-robin for fairness reasons.

Almost nobody does this, which is why routing debates in sales organizations are conducted entirely on the basis of who feels aggrieved. The debate is winnable with one extra field.

The uncomfortable finding you should be prepared for: sometimes the rules are worse than random, usually because they encode an org chart from two years ago. Territories drift, reps develop new specialisms, and a rule written when the company sold into one market keeps routing confidently after the market has changed. A router nobody audits is a set of confident assertions about a world that has moved.

Segment, then relationship, then capacity, then coin flip. Assign in under a second. Write down why.

— DealArena Team