Lead routing decides which person gets which inbound record, and how quickly. It is a small piece of plumbing with effects out of proportion to its size, because it sits between everything marketing spends and anybody having a conversation.
It also attracts rules the way a cupboard attracts cables. Each one was justified at the time. Together they form a system nobody has read end to end, which quietly makes decisions the scoring model was supposed to make.
Speed is the variable with real evidence behind it
A 2011 study in Harvard Business Review by James Oldroyd, Kristina McElheran and David Elkington audited 2,241 US companies on how fast they answered web-generated leads. Firms that tried to make contact within an hour were nearly seven times as likely to qualify the lead as those that tried an hour later. Twenty-three per cent of the companies never responded at all.
The study is fifteen years old and the tooling has changed since. Treat the direction as the finding and the multiple as one audit of one sample rather than a constant. What makes it worth building around anyway is that response time is the only part of this that is purely an engineering decision. Nothing about the buyer, the product or the market has to change for a lead to reach somebody in four minutes instead of four hours.
Round robin is a fairness mechanism
Even distribution is a compensation and morale property. It keeps territories balanced and arguments short, and those are real goods. What it is not is a conversion mechanism, and it is worth being clear about which one you bought.
Round robin asserts that every rep is interchangeable for every lead. If you have built a scoring model that says some accounts are materially better than others, the routing layer has just thrown that conclusion away one step after you reached it. Either the score should steer assignment or the score is decoration.
The inputs worth routing on
| Input | What it buys | Where it breaks |
|---|---|---|
| Territory | Unambiguous ownership and very few disputes. | Remote buyers, holding companies and multinationals, where the address is not where the decision happens. |
| Existing account ownership | Stops two reps calling the same company in the same week. | Entirely dependent on lead-to-account matching; inconsistent company names defeat it silently. |
| Fit score | Sends the best accounts to the people most likely to close them. | Only as defensible as the score, and reps will test whether it is. |
| Availability | Speed, which is the input with the best evidence. | It overrides the other three, and most of the time it should. |
The ordering is the design. A rule set that checks territory, then ownership, then score, then availability will route correctly and slowly. One that checks availability first will route fast and occasionally wrong. Which trade you want is a real decision, and most teams have never made it explicitly.
The rules you can probably delete
Routing engines are almost always first-match-wins, which means a rule placed below a broader one never fires at all. Half a rule set can be dead without anyone noticing, because a dead rule produces no error and no complaint.
- Rules naming a person who has left. These usually survive because the reassignment was done at the record level and never at the rule level.
- Rules that fired zero times last quarter. Count the firings before you argue about the logic. A rule that has never fired is not protecting anything.
- Rules shadowed by a broader rule above them. Same symptom, different cause, and the fix is ordering rather than deletion.
- Rules encoding a one-off exception from a campaign that ended. The campaign has a date. The rule does not.
Do this before adding anything. A rule set small enough to read is a rule set somebody can be responsible for.
The unrouted queue is where leads die
Every routing system needs a documented answer to three questions: what happens when no rule matches, what happens when the chosen owner is away, and what happens when lead-to-account matching fails. The usual answer to all three is that the record sits where it landed and nobody finds out.
Instrument that queue and alert on the age of the oldest item rather than the size of the queue. Volume is noisy and seasonal. Age is unambiguous, and a record sitting unrouted for two days is a specific failure with a specific owner.
When not to automate it
Under a handful of inbound records a week, route by hand. A person reading every one learns the patterns that the rules should eventually encode, and at that volume they are faster than any build.
Before lead-to-account matching is reliable. Automated routing on bad matching distributes errors at speed and with an audit trail that looks authoritative.
While the underlying fields are still ungoverned. Routing reads whatever the CRM says, so the write path has to be settled first or the rules are reacting to formatting rather than to facts.
Questions people ask
What is lead routing?
The logic that decides which person receives an inbound lead and how fast. It is plumbing, and it quietly makes prioritisation decisions whether or not anyone intended it to.
Is round robin a good default?
It is a good fairness default. It is not a conversion strategy, and it discards whatever your scoring model concluded one step earlier.
How fast does routing need to be?
Fast enough that a human can make contact inside the hour, which is the window the Harvard Business Review audit found mattered. Minutes of routing latency are cheap to remove.
How many routing rules is too many?
More than one person can hold in their head. Past that, count how often each rule fires, because a rule set grows by addition and nobody is assigned to the subtraction.
What should happen to a lead no rule matches?
It should go to a named person and raise an alert based on how long it has waited. The common alternative is that it sits in place and nobody learns it existed.
Sources
- James B. Oldroyd, Kristina McElheran and David Elkington, The Short Life of Online Sales Leads, Harvard Business Review, March 2011. Audit of 2,241 US companies; firms attempting contact within an hour were nearly seven times as likely to qualify a lead as those attempting an hour later, and 23 per cent never responded.
Or skip the search
Want the person who owns it after it ships?
If the work is a standing obligation rather than a bounded project, tell me which system you want built first. You get example profiles and a call with your pick before anything starts.
Placement runs through AYTalent. Engineers are employed and trained on the bench before they start with you.