“Our leads go cold before the sales team calls them” — what to do in the next 24 hours and 7 days

When leads go cold before sales calls them, the delay usually sits in routing, ownership or after-hours coverage, not rep effort. In Pied Piper’s May 2026 test of 31 large US auto dealer groups, group scores ranged from 54 to 93 out of 100, and on average only 50% of leads got both a written answer and a phone call within 30 minutes.

At a glance

• Next 24 hours: pull 14 days of leads with the timestamp of the first human call, submit five test leads of your own, and call every untouched lead under seven days old, youngest first.
• Next 7 days: fix routing, define the clock, cover after-hours, respond on two channels, retest on day 7.
• The bar to compare against: the top group in Pied Piper’s study phoned 87% of leads within 15 minutes and answered in writing and by phone within 30 minutes 91% of the time.
• Not urgent yet: one slow week after a campaign launch. Urgent: three straight weeks with a 90th-percentile first call over 24 hours, or more than 5% of leads never called.

Why are our leads going cold before sales calls them?

At enterprise volume, a lead rarely waits on a lazy rep. It waits in a round-robin that still includes someone on leave, in a territory rule that matches no owner, behind a lead-scoring threshold it has not reached yet, or in a form-to-CRM sync that runs hourly. None of these shows up as a missed call, because nobody was asked to make one.

The second mechanism is activity that never reached the customer. An automated “thanks for your inquiry” email gets logged as a touch, and the lead drops off the uncontacted report. Pied Piper describes the same kind of failure in its 2026 Internet Lead Effectiveness dealer group study. Where AI and CRM, email, text and phone systems hand off to each other, “the system may claim activity occurred while the customer still receives no useful response.” It also found that at dealerships relying heavily on AI, the typical response score dropped an average of 9 points when an inquiry needed human involvement.

A CRM that records an auto-reply as “contacted” will report a lead-response problem as solved while the lead goes cold. Pied Piper is a vendor that sells this kind of measurement to dealers and manufacturers, so its study is also marketing for that service. Its method is disclosed: 2,414 inquiries to every dealership in 31 US groups, each with a new name, email and local phone number, sent in business hours and scored over 24 hours. Auto retail is consumer retail, not B2B sales, and car shoppers are not enterprise buyers. What carries over is the spread: same leads, similar organizations, very different results.

Is it as urgent as it feels?

Sometimes not. A launch week that doubles inbound volume, a holiday, or one vacant territory will stretch response times for a week and then recover. Do not restructure a sales floor on seven days of data.

Our working rule, which is a decision rule and not a published benchmark: it is a trend worth acting on this week when either of these holds for three consecutive weeks.

• The 90th-percentile time to first human call is over 24 hours. Use the 90th percentile, not the average, which hides the one lead in ten that waits a day.
• More than 5% of leads have no human call at all. For comparison, Pied Piper’s March 2026 powersports release puts the auto industry average for web leads that got no response of any kind (email, text or call) within 24 hours at 5%. That is a looser test than ours: the same release puts the auto average for phoning the customer at all at 74%, so far more than 5% of those leads got no call. In the dealer group study, the best groups were under 1% and the worst were over 10%.

A slow week is volume; a slow 90th percentile three weeks running is a system, and systems do not fix themselves.

How fast is fast enough? The cold-lead ladder

The cold-lead ladder sorts every uncalled lead by age and gives each band one action. The 15- and 30-minute bars are what Pied Piper measured the best-run US dealer group doing. The actions and attempt counts are our operating rule, a starting point rather than research.

The cold-lead ladder’s rule is to work the backlog youngest first, because a two-hour-old lead can still be saved today and a twenty-day-old lead will be no colder tomorrow.

What to do in the next 24 hours

None of this costs money or needs a vendor.

1. Export 14 days of leads with three fields: created timestamp, timestamp of the first outbound call (not first “activity”), and owner. Leave out auto-replies and sequence emails.
2. Compute three numbers: median time to first human call, 90th-percentile time to first human call, and the count with no call. Then split all three by lead source, territory and hour of arrival. The overall median usually hides a problem confined to one of them.
3. Submit five test leads yourself through your real forms: two in business hours, two in the evening, one on a weekend, each with a different name, email and phone number. Log what arrives, where and when: Pied Piper’s method in miniature.
4. Work today’s backlog down the ladder, starting with leads under 24 hours old.
5. Name one person as owner of the unassigned queue until the routing is fixed. Leads with no owner are the ones that never get a call.

What to do in the next 7 days

1. Days 2–3: fix each routing failure the test leads exposed. Take people on leave out of the round-robin, add a catch-all owner for unmatched territories, and lift scoring holds on inbound hand-raisers. The standard: every lead has a named human owner within 5 minutes, and an alert fires on any lead still unowned at 15 minutes.
2. Day 3: write down when the clock starts and stops. It starts at form submission, not at CRM creation, and stops at a live call attempt. The clauses are laid out in how to set up a speed-to-lead SLA.
3. Day 4: decide on after-hours coverage. Roster a shift, automate the first response, or tell the lead on the confirmation page when a call will come. Any of the three is defensible; silence is not.
4. Day 5: make two channels the standard. In Pied Piper’s study, 87% of dealerships in the top five groups responded on multiple channels, against 44% in the bottom seven. The top five offered a specific appointment time 60% of the time, against 24% for the bottom seven.
5. Day 6: run the ladder’s 1–7 day and 8–30 day rows.
6. Day 7: submit the same five test leads at the same hours and re-run the export. Compare the 90th percentile and the no-call count with day 1.

To stop it recurring, keep sending test leads weekly; Pied Piper reports that three of its top five dealer groups use ongoing measurement, which is the service it sells. Once first calls happen, the next failure point is the attempts that follow, covered in why sales teams stop following up leads.

Worked example: the cold-lead ledger at 2,000 leads a month

Hypothetical inputs; substitute your own export.

• 2,000 inbound leads a month, so a 14-day export holds 2,000 × 14 ÷ 30 ≈ 933 leads.
• 30% arrived outside Monday–Friday, 9 a.m.–5 p.m.: 280 leads.
• 6% were never called: 56 leads. 14% were first called after 24 hours: 131 leads.

Rescue workload. 56 + 131 = 187 leads × 3 attempts × 5 minutes = 2,805 minutes, or about 47 hours. Spread across 8 reps, that is about 5.8 hours each this week. That is a backlog, not a hiring case.

The after-hours ceiling. A 9-to-5 weekday roster covers 40 of the week’s 168 hours, or 23.8%. A lead submitted at 5:01 p.m. on Friday cannot get a call before 9 a.m. Monday, which is 64 hours later and already in the 1–7 day band. With 30% of leads arriving outside staffed hours, no amount of rep discipline can put more than 70% of leads inside the 15-minute band until coverage changes. Pied Piper tested only in business hours, so its figures are a best case.

Staffed hours are not the same as staffed leads. Most leads arrive in business hours, which is why a roster covering 23.8% of the week’s hours still covers 70% of the leads in this example. Use your own arrival-by-hour split, not the hours ratio, to size after-hours coverage.

What the fix costs to run yourself, and when it is not a sales problem

In-house, the first ledger takes two to four analyst hours and needs a CRM call-log field, not just an activity count. Test leads take 30 minutes a week plus a pool of spare emails and phone numbers; vary names, sources and timing, because reps learn to spot them. Routing takes one to three days of CRM admin time. After-hours coverage is the expensive part: a second shift, an automated first response (what speed-to-lead automation involves in a US business), or accepting the ceiling above.

The two published effect sizes disagree, and they measure different things. Pied Piper says that, historically, dealers who move from an ILE score under 40 to over 80 “sell 50% more units from the same quantity of internet leads”. That score covers speed, channel and content together, and it is the vendor’s own historical claim.

Separately, in our own client work we typically see fixing speed to lead on its own lift conversion by about 3x. That is our operator observation, not a controlled study or a published dataset, and it is not a guarantee. It is not comparable with Pied Piper’s figure, which measures something different. Outsourcing the first call at corporate scale is covered in AI appointment setting for corporate sales teams.

Three situations are not sales problems. If more than a tenth of the phone numbers in the export are invalid, the lead source is the problem; that goes to marketing or the lead vendor. If you plan to re-contact old leads with an autodialer or an artificial or prerecorded voice, consent is a legal question for counsel first (AI calling and the TCPA for US businesses). If one named rep is the problem, that is for HR.

Frequently asked questions

How fast should sales call a new inbound lead?

Within 15 minutes by phone, with a written answer to the question inside 30 minutes. Those are the bars the top-ranked group in Pied Piper’s 2026 dealer group study cleared 87% and 91% of the time, against a 50% group average for the 30-minute measure.

Are leads that are a week old still worth calling?

Yes, but not with a generic script. A lead one to seven days old gets three attempts across seven days on two channels, each naming what the person originally asked and offering a specific time. After 30 days it belongs in a separate dormant-lead program, not the sales response queue.

Why does our CRM say leads were contacted when customers say nobody called?

Usually because an automated email or text was logged as a touch. Pied Piper’s 2026 study warns that where AI, CRM and messaging systems hand off to each other, the system may claim activity occurred while the customer receives no useful response. Measure the first human call, not first activity.

Can we legally call old web leads again in the US?

It depends on how you call and what consent you hold. The TCPA, 47 U.S.C. § 227, restricts calls to cell phones made with an autodialer or an artificial or prerecorded voice without the called party’s prior express consent, and prerecorded-voice calls to residential lines without it. This is general information, not legal advice. Check your consent records with counsel before an automated re-contact campaign.

Should we hire more SDRs to fix slow lead response?

Not until you have measured where the delay sits. If leads wait in routing or arrive after hours, more reps on the day shift change little: a 9-to-5 weekday roster covers 40 of the week’s 168 hours. Hire when the ledger shows leads owned, in hours, and still waiting.