Lead routing is the most consequential automation most companies never look at. It decides, silently and thousands of times a quarter, which human being gets the chance to turn interest into revenue — and unlike a broken email send, a broken routing rule produces no bounce report. The lead just sits there, and the revenue just doesn’t happen.
I covered why slow routing costs so much in the speed-to-lead piece. This one is the practical companion: how routing is typically built in HubSpot, the ways it rots, and the quarterly audit that keeps it honest.
How routing rot happens
Nobody designs bad routing. It accretes. The original build was clean: a few territory rules, a round-robin, an owner-assignment workflow. Then reality arrived, one patch at a time — a new segment for the enterprise push, an exclusion for the partner channel, a temporary rule for an event that ended two years ago, a special case for a rep who’s since been promoted. Each patch was rational. The stack of patches is now a system no single person fully understands, and systems nobody understands fail in the four classic ways: unmatched leads pooling silently, ghost reps in rotations, overlapping rules bouncing ownership, and notifications that fire into the void.
The quarterly routing audit (60–90 minutes)
1. Map what exists (20 min). List every mechanism that can assign an owner: workflows with owner-assignment actions, round-robin/rotation settings, form-level assignment, integration writes (Salesforce assignment rules count if you’re synced), and any manual triage habit. Most instances have more assignment paths than anyone expects — and every path is a place rules can disagree.
2. Find the unrouted (10 min). Active contacts/companies created in the last 90 days with no owner or a generic owner. Target: zero. Reality: rarely zero. Every record here fell through every rule you have — read a sample and you’ll usually find a whole segment the rules never contemplated (new geo, new form, new product line).
3. Verify the humans (10 min). Every rotation and every hard-coded owner in every workflow, checked against the actual current team. Departed reps, people on leave, changed territories. This is the least glamorous check and the most reliably productive one.
4. Hunt overlaps (15 min). For each pair of active assignment workflows, ask: can one record satisfy both? Pull records with 3+ owner changes in their first week — that thrash pattern is overlapping rules fighting, and every bounce resets your response clock.
5. Test the full chain (15 min). Submit test leads that match your main segments — and one designed to match nothing. Time each step: assignment latency, notification delivery, task creation. The catch-all test lead is the important one: it should route to a named human and fire an alert, because unmatched leads should page someone, not pool.
6. Read the exhaust (10 min). Workflow error logs and any assignment-related sync errors. Cluster, fix the top cluster, move on.
Design principles for the rebuild
If the audit reveals more rot than patching can fix, rebuild on four principles:
- Fewer, broader rules. Ten rules a team understands beat forty rules that cover 2% more cases. Complexity is where rot breeds.
- Always a catch-all. The final rule matches everything unmatched and routes to a named owner with an alert. No silent pools, ever.
- Route on few, reliable fields. Every field a rule reads is a dependency on data hygiene — route on country and company size (usually reliable) rather than a self-reported field that’s blank 30% of the time.
- Date-stamp temporary rules. Any rule built for a campaign or event gets an expiry review date in its name. “Temporary” is how permanent rot begins.
A routing audit almost always turns up a pool of leads matching no rule at all — sitting unassigned and unworked, invisible because no dashboard counts a lead that landed nowhere. Of the failure types, unrouted-because-no-rule usually dominates, followed closely by rotations that still include reps who’ve left. The speed-to-lead effect of fixing it is concentrated in the tail, not the average: the leads that were waiting indefinitely start getting worked in minutes, while the ones that already routed cleanly don’t change. That tail is where the lost pipeline was hiding.
Put it in the calendar
The audit above belongs in the same quarterly ritual as the MQL definition review and the SLA check — they’re one hygiene practice viewed from three angles, and they decay at the same rate. Ninety minutes a quarter is the entire cost of never again discovering, six months too late, that an entire segment of your demand has been routing to nobody.
—
Routing is part of lens two of the Pipeline Leak Audit — audited against your live instance with evidence. Or bring your unrouted-lead count to a free health check and we’ll read it together.