Of all the numbers in a B2B funnel, speed to lead — the time between a lead raising their hand and a human responding — has the least glamorous name and some of the highest leverage. It’s also the number most companies measure wrong, or not at all.
Why it matters more than almost anything else you A/B test
The research on this has been consistent for over a decade: contact attempts within the first minutes of an enquiry convert at multiples of attempts made hours later, and the drop-off is brutal — interest decays fast, competitors respond faster than you think, and a demo request left overnight is often a demo request answered by someone else.
You don’t need to trust the industry studies, though. Run the maths on your own data: segment last quarter’s inbound leads by response time (under 1 hour, 1–24 hours, over 24 hours) and compare conversion to opportunity across the three bands. In nearly every dataset I’ve looked at, the under-one-hour band converts at two to four times the over-24-hour band. That gap, multiplied by the share of your leads landing in the slow band, is the size of this leak in your pipeline — in currency.
Measure it honestly: median AND 90th percentile
Most teams that track speed to lead track the average, and the average lies. A handful of instant automated responses drags the mean down while a long tail of leads waits days. Two numbers tell the truth:
- **Median time to first human touch** — not the auto-reply email; the first genuine sales action.
- 90th percentile — how long the unluckiest tenth of your leads wait. This is where the routing failures hide.
A healthy funnel: median under an hour on hot inbound (demo requests, pricing pages), 90th percentile under a working day. If your 90th percentile is measured in days, you don’t have a rep effort problem — you almost certainly have a routing problem.
The four routing failures behind slow response
Routing logic rots. It’s built once, patched often, and audited never — and it fails in four predictable ways:
1. The unmatched lead. Rules were written for known segments; anything that fits none of them falls into an unassigned queue nobody watches. Check: filter for leads with no owner, or owned by a generic user, created in the last 90 days. This number should be zero. It rarely is.
2. The ghost in the round-robin. Reps leave; their rotation slot doesn’t. Leads assigned to departed or on-leave owners sit untouched until someone notices. Check the active membership of every rotation against the actual team.
3. The double-match bounce. A lead matches two overlapping rules and gets reassigned back and forth, resetting SLA timers each time. Look for records with multiple owner changes in their first 48 hours.
4. The silent notification failure. The routing worked — the rep just never found out. Notification workflows break quietly when email templates change, Slack integrations are re-authed, or a task queue gets muted. Test the full chain end to end: submit a test lead, time how long until a human sees it.
Measured honestly — from form submission to a human actually seeing the lead, not from the CRM’s created-date — the 90th percentile is often startlingly worse than the median everyone quotes. The median looks fine because most leads route cleanly; it’s the tail that hurts, and the tail is usually one specific failure rather than general slowness. The silent notification break is the most common culprit: routing works, the record is assigned, and the rep is simply never told. Reconnect that one link and the tail collapses back toward the median — no new tooling, just a chain that was quietly severed.
The fix sequence
1. Instrument first. Timestamp lead creation and first human touch; build the median/90th percentile report. You can’t manage what you argue about.
1. Drain the unassigned queue and add a catch-all rule: anything unmatched routes to a named human, with an alert. Unmatched leads should page someone, not pool silently.
1. Audit the rules quarterly. Fifteen minutes: active rotations vs. actual team, overlap check, test lead end to end. Put it in the calendar next to the MQL definition review — they rot at the same speed.
1. Only then talk about rep behaviour. SLAs on humans are unenforceable while the system is misrouting. Fix the plumbing, then set the SLA.
Speed to lead is rare among funnel metrics: no budget increase, no new tooling, no creative debate — just an audit of rules you already own. It’s usually the fastest pipeline win available.
—
Routing and speed-to-lead are lens two of the Pipeline Leak Audit. Want to know if it’s your leak? Book a free health check — bring your 90th percentile if you have it.