A plumber in Houston told me he lost $14,000 in a single month because his routing sent after-hours emergency calls to a buyer who didn't have weekend coverage. The calls connected — technically. To voicemail. Sixty-three emergency plumbing leads, gone.
His routing logic looked fine on paper. "If caller is in Texas, route to Texas buyer." It was. The Texas buyer just wasn't answering on Saturdays.
That's the problem with call routing. The obvious setups miss the obvious edge cases. And in pay-per-call, edge cases aren't edge cases — they're 15-20% of your volume, quietly bleeding money every week.
I've spent three years auditing routing setups for operators across home services, legal intake, insurance, and Medicare. The mistakes repeat. The fixes repeat too. Here are the nine call routing practices that actually move the needle — not theory, not "best practices" copy-pasted from a help doc, but the specific rules that separate campaigns making money from campaigns wondering where the margin went. For broader context on pay-per-call economics, see the 2026 pay-per-call benchmarks report.
1. Set Ring Timeouts at 20-25 Seconds — No More
This sounds basic. It isn't — because most operators either inherit platform defaults (often 30-45 seconds) or set timeouts based on what feels right rather than what data shows.
Here's the problem with long ring times: caller abandonment increases roughly 1% for every 2 seconds past the 20-second mark. At 30 seconds, you've lost 5% of callers. At 45 seconds — the default on some platforms — you're approaching 12-15% abandonment before the call even connects.
The counter-argument: "But my buyers need time to pick up." If your buyers consistently need 30+ seconds, the problem isn't your timeout. It's their staffing. A buyer who can't answer in 20 seconds is a buyer who shouldn't be your priority route.
Set timeouts at 20-25 seconds. Build escalation rules for what happens when that timeout fires. Don't let callers ring into oblivion hoping someone eventually picks up.
I made this mistake on an early HVAC campaign — 40-second timeouts because the buyer asked for them. Took me three weeks to notice. Three weeks. Lost 18% of volume to abandonment before I figured out what was happening. The buyer was happy with their comfortable ring time. My margin was not.
For more on IVR and ring-time tradeoffs, see the IVR abandonment rate study.
2. Match Buyer Concurrency to Actual Intake Capacity
Every buyer will tell you they can handle more calls than they actually can. It's not lying — it's optimism. The intake person says "I can do 10 calls at once" because they've never actually tested it.
Real-world capacity:
- Solo intake person: 2-3 simultaneous calls before quality drops
- Team of 3: 5-7 calls
- Team of 5: 8-12 calls
- Dedicated call center (10+ seats): 15-25 calls, depending on average handle time
When you exceed a buyer's real capacity, one of two things happens. Calls wait in queue — and callers abandon. Or calls dump to voicemail. Nobody converts. Neither is good.
The fix: start conservative. Route 50% of what the buyer claims they can handle. Monitor answer rates. If answer rate holds above 90% for two weeks, increase by 20%. Repeat until answer rate dips more than 5 points — that's their ceiling.
And for the love of margin, don't route to buyers who are already maxed out. If Buyer A has 3 active calls and capacity for 3, skip to Buyer B. Real-time capacity checks aren't optional. They're the difference between 85% fill rate and 65% fill rate.
3. Layer Geo + Time-of-Day + Day-of-Week
Geography gets all the attention. Time-of-day gets lip service. Day-of-week gets ignored. You need all three.
Geography: Route callers to buyers who actually serve their area. This seems obvious until you audit a campaign routing Miami callers to a buyer with Tampa coverage only. "Close enough" isn't close enough. Service-area mismatches are the single most common routing error I see — and usually the easiest to fix.
For geographic routing setup, see route calls by caller location.
Time-of-day: A call at 9am Monday and a call at 9pm Friday need different routing. Your primary buyer might be perfect for business hours but non-existent after 5pm. Build time-based rules that respect actual buyer availability, not just "business hours."
The time windows that matter:
- 8am-10am local: High pickup, but buyers are ramping up — ease in volume
- 10am-12pm: Peak. Route aggressively.
- 12pm-1pm: Lunch dip. Have backup buyers who stagger lunches.
- 1pm-5pm: Solid, but watch for early closers on Fridays.
- 5pm-8pm: After-hours for most. Route to buyers with evening coverage only.
- 8pm-8am: Night/weekend. Many verticals dead here; some (emergency plumbing, PI intake) are gold.
Day-of-week: Friday afternoon ≠ Tuesday afternoon. Weekend ≠ weekday. Holiday weeks have their own patterns. Build explicit rules, not assumptions.
The plumber I mentioned at the start? His routing had geo covered. It had time-of-day — "business hours." It didn't have day-of-week. Saturdays routed to the same buyer as Tuesdays. Buyer wasn't there on Saturdays. Sixty-three calls to voicemail.
4. Build a Fallback Chain — Every Call Needs Somewhere to Go
Priority routing is great until Priority 1 can't take the call. Then what?
Every routing setup needs a fallback chain:
Priority 1: Primary buyer (best payout, best fit)
Priority 2: Secondary buyer (same region, slightly lower payout)
Priority 3: Overflow buyer (national, 24/7, lower margin)
Priority 4: Voicemail with callback promise
Priority 5: Polite decline (last resort — "We couldn't connect you")
The goal: 0% of calls should hit a dead end. Every call, even at 3am on Christmas, should have a path.
I've audited campaigns where 8-12% of calls hit no-route conditions and just... vanished. Platform counted them as "completed" because the IVR ran. But nobody picked up. Nobody converted. Pure loss.
The fallback chain doesn't have to be profitable at every tier. Priority 3 and 4 might be margin-neutral or even margin-negative. That's fine — you're recovering volume that would otherwise be zero revenue. A voicemail callback that converts at 15% beats a dropped call that converts at 0%.
For the mechanics of ping trees versus direct fallback, see ping tree vs direct buyer routing.
5. Set Per-Buyer Daily and Hourly Caps
Buyers have budgets. Buyers have capacity. Buyers have attention spans.
A buyer who says "send me everything" will — at some point — stop answering. Either they hit their daily budget, their intake team burns out, or they simply lose interest once they've booked enough appointments for the week. If you keep routing to them past that point, you're wasting calls.
Daily caps: Set based on buyer's actual budget divided by your average payout. If a buyer has $2,000/day to spend and calls average $50, cap at 40 calls/day. Leave 10-15% buffer for conversion variance.
Hourly caps: Even buyers with large daily budgets can't handle spike volume. A buyer who wants 60 calls/day probably can't handle 30 of them in a single hour. Spread the volume. I usually set hourly caps at 15-20% of daily cap — so a 60-call buyer gets capped at 9-12 calls/hour.
Cap reset timing: Match the buyer's billing cycle. If they reset budget at midnight Pacific, your caps should reset at midnight Pacific. Misaligned resets create weird gaps — buyer has budget from 12am-3am your time, but your caps already reset at midnight Eastern.
Caps protect you from over-routing. They also protect buyer relationships. Nothing burns a buyer faster than getting 150 calls in a day when they asked for 80. They won't complain — they'll just stop answering and quietly find another source.
(Ask me how I learned this. Actually, don't.)
6. Weight Routing by Conversion Rate, Not Just Payout
The obvious routing logic: route to the highest-paying buyer first.
The better logic: route to the buyer who actually closes.
Here's the math. Buyer A pays $70 per call and closes 12% of leads. Buyer B pays $55 and closes 28%. For every 100 calls:
- Buyer A: 100 calls × $70 = $7,000 revenue. 12 closed deals for the buyer.
- Buyer B: 100 calls × $55 = $5,500 revenue. 28 closed deals for the buyer.
Your revenue is higher with Buyer A. But Buyer A is getting 12 deals from 100 calls — terrible economics on their end. Eventually, Buyer A either demands lower pricing or walks entirely. Meanwhile, Buyer B is thriving.
The play: Prioritize buyers who convert well, even if payout is lower. Your per-call revenue drops, but your buyer retention improves, your traffic sources stay cleaner (high-converting buyers don't complain about lead quality), and your long-term margin is more stable.
I weight routing by close rate first, payout second. A buyer closing at 25% gets priority over a buyer closing at 15%, even if the 15% closer pays 20% more. The relationship lasts longer and causes fewer headaches.
The exception: if you're running real-time bidding where buyers set their own prices, let the auction sort it. Buyers bidding high on calls they can't close will learn quickly. But for direct routing with negotiated payouts, close rate matters more than rate card.
Honestly, the payout chase is a trap. I've watched operators chase an extra $15/call and lose their best buyer in the process. Short-term thinking. Don't do it.
7. Implement AMD and Post-Connect Qualification
Calls that "connect" to voicemail are not calls that connect. Calls that connect to a fax machine, a wrong number, or a disconnected line are not calls.
AMD (answering machine detection) catches voicemail pickups before they count as connects. It's not perfect — false positives happen, and aggressive AMD can accidentally filter real pickups that just have slow greetings. But AMD catches 70-80% of voicemail connects that would otherwise waste your buyers' time.
Post-connect qualification goes further. Did the call reach a human? Did the human engage past 30 seconds? Did they express intent? A 2-minute call where someone says "wrong number" and hangs up is not a qualified connect, even if it technically lasted long enough to bill.
Most platforms track duration. Fewer track engagement. If your platform supports it (VeloCalls' AI conversation intelligence layer does this, as do Ringba and some others), use conversation signals — did the caller state a problem? Did they mention a service need? Did they schedule or express intent to schedule?
This isn't just for billing accuracy. It's for routing optimization. If Buyer A's calls show 40% meaningful engagement and Buyer B's show 65%, you know where to prioritize volume, independent of what the duration-based "qualification" says.
For junk call filtering before calls even hit the IVR, see filter junk calls.
8. Separate Routing Rules by Vertical and Call Type
"HVAC" is not one vertical. Emergency AC repair in Phoenix in July is a different call than routine furnace maintenance in Minneapolis in March. The buyers are different. The payouts are different. The conversion patterns are different.
Build separate routing rules for:
- Service type: Emergency vs. scheduled. Install vs. repair vs. maintenance.
- Ticket size: A $15,000 HVAC replacement has different routing than a $150 tune-up.
- Caller intent: Someone ready to book today vs. someone price-shopping for next month.
The more you segment, the better you can match calls to buyers who actually want them. A buyer specializing in high-ticket replacements doesn't want $150 tune-up calls clogging their queue. A buyer running a 24/7 emergency line doesn't want 9-to-5 appointment requests.
In legal intake, this matters even more. PI auto and mass tort are completely different operations. A PI firm tooled for car accident intake can't handle mesothelioma calls — the qualification criteria, the case value, the follow-up process are all different. Route them to different buyers or you'll burn both relationships. For legal-specific routing considerations, see our legal pay-per-call guide.
For vertical-specific routing nuances, the home services pay-per-call guide covers HVAC, plumbing, and roofing in detail.
9. Monitor Fill Rate Weekly and Diagnose Gaps
Fill rate — the percentage of inbound calls that successfully route to a buyer who answers — is your canary in the coal mine.
Healthy fill rate: 88-95%, depending on vertical and time mix.
If fill rate drops below 85%, something is wrong. Common causes:
- Buyer availability gaps: Check if specific time windows or days are underperforming
- Capacity maxed: Buyers hitting caps before the day ends
- Routing logic errors: Rules that accidentally exclude valid calls
- Spam flagging: Your tracking numbers getting filtered by carriers (see fix low answer rates for this rabbit hole)
- Buyer churn: A buyer stopped answering but you didn't notice
Track fill rate by hour of day. Find the gaps. Track by day of week — Friday afternoon is always worse, but how much worse? Track by buyer. Who's underperforming? Track by geo region and traffic source. Are some publishers sending calls you can't match anywhere?
Weekly review for the first 90 days. Monthly after that. The operators who monitor fill rate catch problems early. The ones who don't find out when margin is already gone.
If you're running paid search traffic into these calls, ClickzProtect handles the click-fraud side — bad traffic generates bad calls that your routing can't save.
Honorable Mentions
A few practices that didn't make the main list but matter for specific situations:
Sticky routing for callback scenarios. If a caller reaches Buyer A on the first call but needs a callback, route the callback to Buyer A — not whoever's next in priority. Continuity matters. Some platforms call this "call memory" or "session persistence."
A/B testing routing variations. Split traffic 50/50 between two routing configurations and measure fill rate, conversion, and margin. Most operators guess which setup is better. Test instead. (I say this having guessed wrong on routing changes more times than I'd like to admit. My gut is not as smart as it thinks it is.)
Fraud scoring in the routing layer. Calls from known VOIP ranges, repeat callers, datacenter IPs, and other fraud signals should either skip bidding entirely or route to a quarantine flow. Don't let fraudulent calls burn your premium buyers' time and trust.
Quick Verdict
If you implement one thing from this list, make it the fallback chain. Every call needs somewhere to go. A single dead-end path that drops 10% of calls will cost you more than any other routing inefficiency.
Second priority: time-of-day and day-of-week rules. Geography gets attention; timing gets ignored. Fix that.
Third: weekly fill-rate monitoring. You can't optimize what you don't measure.
The routing rules that work aren't the fancy ones. They're the ones that handle the edge cases — the Saturday call, the 7pm emergency, the buyer who's already at capacity. Get the basics right, monitor the gaps, and the margin follows.
And if you're still routing to buyers who asked for 30-second ring times? Fix that first. I'm begging you.
Frequently Asked Questions
What's the ideal ring timeout for pay-per-call routing?
20-25 seconds. Shorter than that and you're not giving buyers a fair chance to answer. Longer than that and caller abandonment spikes — industry data shows abandonment increases roughly 1% for every 2 seconds of ring time past 20 seconds. If your buyers consistently need 30+ seconds to pick up, the problem isn't your timeout — it's their staffing.
How many concurrent calls should I route to a single buyer?
Match their actual intake capacity, not what they claim. A solo intake person can handle maybe 2-3 simultaneous calls before quality drops. A team of 5 can handle 8-12. Ask buyers directly, then test by monitoring their answer rates as you increase volume. When answer rate dips more than 5 points, you've found their real ceiling.
Should I prioritize routing by buyer payout or buyer conversion rate?
Conversion rate. A buyer paying $60 but closing at 25% is worth more than a buyer paying $80 but closing at 12% — because you'll burn through volume faster and damage your traffic source relationships routing to low-converters. Weight routing by close rate first, then use payout as a tiebreaker.
How often should I review and adjust my routing rules?
Weekly for the first 90 days, monthly after that. Buyer performance shifts — seasonality, staffing changes, and vertical economics all fluctuate. The routing rules that worked in Q1 might quietly bleed margin by Q3. Set a calendar reminder. Don't trust that nothing changed.
Try VeloCalls for Your Vertical
Pay-per-call platform built for HVAC, plumbing, roofing, PI lawyers, Medicare brokers, and insurance. Smart routing, real-time bidding, visual IVR builder, AI conversation intelligence (transcription, sentiment, summarization, AMD). Per-minute pricing — Managed starts at 4¢/min, BYOC at 2¢/min, both drop as you scale.