An operator pinged me last Tuesday asking why his EPC dropped from $47 to $31 in two weeks. I asked about his billable percentage. He said 68%. I asked about last month. "Around 70%."
That's not the problem, then.
What about answer rate? "Uh... somewhere in the 80s?"
We dug in. His answer rate had collapsed from 91% to 76%. His buyers hadn't changed their payouts—they just weren't getting connected often enough to pay. Fifteen percent of his traffic was ringing out to voicemail during business hours. Staffing gap. A team member quit, nobody backfilled, and he'd been watching EPC without knowing why it cratered.
The metrics told the whole story. He just wasn't reading them. And honestly? I've been there. Back when I started running campaigns, I fixated on payout per call like it was the only number that mattered. Took six months and a lot of burned money to realize the operational metrics—answer rate, billable percentage, fill rate—are what keep payout per call healthy.
This is the fourth glossary in the family. The pay-per-call glossary covers marketplace terms. The call tracking glossary covers attribution. The IVR routing glossary covers call flows. This one covers the metrics layer—the numbers you stare at in dashboards trying to figure out why last week was great and this week isn't.
Revenue Metrics
1. RPC (Revenue Per Call)
Total revenue divided by total calls. Send 1,000 calls, earn $38,500, your RPC is $38.50.
Simple. Maybe too simple. RPC doesn't distinguish between a 45-second hang-up and a 12-minute qualified lead. Both count as "one call." For campaigns with duration-based payouts, RPC can mislead—your average looks fine while short calls drag down revenue.
Still, RPC is the number buyers and publishers communicate in. "What's your RPC on HVAC?" is a question you'll answer every week in this industry.
2. RPM (Revenue Per Minute)
Total revenue divided by total connected minutes. $38,500 from 2,100 connected minutes = $18.33 RPM.
RPM matters when call length correlates with payout. Legal intake is the classic example—a 2-minute call that doesn't qualify pays nothing, a 9-minute call with a signed retainer pays $400. In those verticals, RPM tells you more than RPC ever could.
For fixed-payout campaigns (flat $45 per qualified call regardless of duration), RPM is noise. Use RPC.
3. EPC (Earnings Per Call)
Publisher-side metric. Same math as RPC, but framed from the traffic source's perspective. An affiliate sends 500 calls, earns $21,000, EPC is $42.
Networks and buyers often quote payouts; publishers care about EPC because it factors in fill rate, billable percentage, and actual earnings—not just what the buyer says they'll pay.
4. Cost Per Call (CPC)
What you spend to generate each call. $3,200 in ad spend producing 200 calls = $16 CPC.
Don't confuse this with "payout per call"—CPC is your acquisition cost, not your revenue. The gap between CPC and RPC is your margin. $16 CPC and $42 RPC? You're making $26 per call. $16 CPC and $18 RPC? Find better traffic or renegotiate payouts.
For monitoring ad spend efficiency, ClickzProtect catches fraudulent clicks that inflate CPC without generating real callers.
5. ROAS (Return on Ad Spend)
Revenue divided by ad spend. $38,500 revenue from $12,000 spend = 3.2x ROAS.
ROAS is the number that matters for media buyers. Below 2x, you're probably underwater after overhead. Above 4x, you're either crushing it or your attribution is broken. (Usually the second one.) Most profitable pay-per-call campaigns run 2.5-4x ROAS once they stabilize.
Connection Metrics
6. Answer Rate
Percentage of calls answered by an agent. 850 answered out of 1,000 = 85% answer rate.
This is the operational metric that kills campaigns quietly. A 90% answer rate versus 80% is a 10% revenue gap—same traffic, same payouts, just better staffing and routing. Track it by hour, by day, by destination. Most operators find patterns: the Monday morning flood that overwhelms agents, the Friday afternoon ghost shift when nobody picks up.
If your answer rate dips below 85%, something's broken. Investigate before optimizing anything else.
7. Connect Rate
Calls that successfully reach a buyer destination divided by calls attempted. Similar to answer rate but focused on the routing layer, not agent availability.
A call can "connect" (reach the destination) but not be "answered" (picked up by a human). Connect rate measures whether your routing is working—are calls reaching the right place? Answer rate measures whether someone's there to pick up.
If connect rate is low but answer rate is high, your routing rules are the problem. If connect rate is high but answer rate is low, your buyers are understaffed.
8. Abandonment Rate
Callers who hang up before reaching an agent. Often measured at the IVR or queue stage.
The IVR abandonment study found abandonment spikes after 4 menu prompts and climbs steadily after 30 seconds of hold time. Industry baseline is 10-15% abandonment; emergency services run lower (callers are desperate), considered purchases run higher (callers are browsing).
Abandonment is money walking away. A 20% abandonment rate on $50 calls is $10 lost per hundred calls—before you even talk to anyone.
9. Average Wait Time
How long callers sit in queue before reaching an agent. Also called Average Speed to Answer (ASA).
Under 30 seconds is excellent. Under 60 seconds is acceptable. Over 90 seconds and you're losing callers to abandonment. Track the 90th percentile, not just the average—your worst wait times are killing more calls than your average suggests.
10. Ring Time
Seconds between the call arriving at a destination and the agent picking up. Relevant for hunt groups and ring trees where multiple agents might be tried.
Long ring times (15+ seconds per agent) compound. If your routing tries three agents sequentially at 15 seconds each, callers wait 45 seconds minimum—assuming someone answers. Tighten ring timeouts or switch to simultaneous ring.
Quality Metrics
11. Billable Percentage
Calls that meet payout criteria divided by total calls. If your buyer requires 90-second minimum duration, billable percentage tells you what portion of calls cross that threshold.
200 calls, 156 billable = 78% billable percentage. The other 22% generated zero revenue.
Low billable percentage usually means one of three things: bad traffic (callers with no intent), aggressive duration thresholds, or IVR that's scaring people off. Diagnose before blaming traffic—I've seen operators blame publishers for "junk calls" when their own IVR was driving 30% of callers to hang up in frustration.
12. Qualified Rate
Percentage of calls that meet qualification criteria beyond duration—homeowner status, service area, immediate need, budget. Usually determined by agent disposition codes or AI scoring.
Different from billable percentage: a call can be billable (90+ seconds) but not qualified (caller is a renter, out of area, or just price shopping). Qualified rate tells you about lead quality, not just call duration.
Buyers often negotiate payouts based on qualified rate. "Your qualified rate dropped to 34% so we're cutting payout by $8" is a conversation you don't want to have.
13. Fill Rate
Percentage of calls successfully routed to a buyer. 1,000 calls arrive, 920 connect to buyers, fill rate is 92%. The other 8% hit no available destination—buyers at capacity, routing rules excluding them, or simply nobody bidding.
Fill rate is the marketplace health metric. High fill = buyers want your traffic. Low fill = something's wrong with your inventory, your floor price, or buyer capacity. For deeper analysis on which traffic sources drive the best fill, JustAnalytics tracks session behavior alongside call outcomes.
14. Conversion Rate
Calls that achieve the desired business outcome divided by total connected calls. Definition varies by vertical: appointments booked, policies sold, retainers signed, estimates scheduled.
This is the metric buyers care about most. They might tolerate 80% answer rate if your conversion rate is 22%. They won't tolerate 95% answer rate if your conversion rate is 6%.
Conversion rate is where agent skill shows up. Two sources sending identical traffic to the same buyer can have wildly different conversion rates based purely on which agents handle the calls. Track it by agent, not just by source.
15. Repeat Caller Rate
Percentage of calls from numbers that have called before within a defined window. Can signal quality issues (callers calling back because first call didn't resolve their need) or suppression failures (junk numbers not getting blocked).
Some repeat calls are legitimate—complex services with multi-touch sales cycles. But a 25% repeat rate in emergency plumbing? Those are unresolved first calls or tire-kickers. Investigate.
Duration Metrics
16. Average Handle Time (AHT)
Total time agents spend on calls divided by number of calls. Includes talk time plus after-call work (notes, disposition codes, CRM updates).
AHT varies wildly by vertical. HVAC scheduling might average 4 minutes. Legal intake can run 12-15 minutes. Medicare Advantage during AEP pushes 20+ minutes.
Lower AHT isn't always better. Rushing calls tanks conversion. But AHT that's twice your competitors' suggests process problems or excessive after-call work.
17. Average Talk Time
Time spent actually talking to the caller, excluding hold time and after-call work. The core component of AHT.
If AHT is 9 minutes but average talk time is 5 minutes, you're spending 4 minutes on hold and wrap-up. That's worth optimizing.
18. Billable Duration
The portion of call time that counts toward payout. A buyer with a 90-second minimum only pays for minutes after that threshold.
A 5-minute call doesn't mean 5 minutes of billable duration. With a 90-second gate, you've got 3.5 billable minutes. VeloCalls charges per minute at rates that step down as volume scales—4¢/min Managed, 2¢/min BYOC at the Starter tier—so understanding billable duration affects both your payout and your platform costs.
19. Duration Distribution
The spread of call lengths across your traffic. What percentage of calls are under 60 seconds? Between 2-5 minutes? Over 10 minutes?
Duration distribution reveals traffic character better than averages. An average of 4 minutes could mean: (a) most calls are 3-5 minutes, or (b) half the calls are 45 seconds and half are 7 minutes. Those are very different traffic profiles requiring different optimization strategies.
Efficiency Metrics
20. Agent Utilization
Percentage of logged-in time that agents spend on calls or after-call work.
70-80% is healthy. Below 60% means overstaffing. Above 85% means agents have no breathing room—quality suffers, burnout rises, turnover follows. This is the number workforce managers obsess over.
21. Calls Per Hour
Volume metric tracking how many calls each agent handles hourly. Depends on AHT—if your calls average 6 minutes, max theoretical throughput is 10 calls per hour (ignoring after-call work).
Useful for capacity planning. You know traffic patterns. You know calls per hour per agent. Math tells you how many agents each hour needs.
22. Cost Per Acquisition (CPA)
What it costs to acquire a converted customer—not just a call, but a closed deal. Total spend divided by closed deals.
CPA is the reality check. You might have great RPC and solid ROAS, but if CPA exceeds customer lifetime value, you're still losing money. Tracking CPA requires integrating call data with downstream conversion data—CRM, signed contracts, closed jobs. Worth the effort. For insights into connecting call data with CRM outcomes, check the HubSpot integration guide.
23. Revenue Per Agent Hour
Total revenue generated divided by total agent hours worked. Captures both call volume and conversion effectiveness in one number.
$4,200 generated across 60 agent hours = $70 revenue per agent hour. If you're paying agents $20/hour loaded, you're generating 3.5x their cost. Below 2.5x gets uncomfortable. Above 4x means you're probably understaffed (or your agents are exceptional—either way, add headcount).
Honorable Mentions
Service Level — Percentage of calls answered within a target time (e.g., 80% of calls answered in 20 seconds). A contact center standard that's less common in pay-per-call but shows up in enterprise buyer requirements.
Transfer Rate — Percentage of calls transferred to another agent or department. High transfer rates signal routing problems or mismatched IVR menus. Every transfer is friction.
Hold Time — How long callers spend on hold during a call (not pre-connect wait time). Excessive hold kills customer experience. If agents are constantly putting callers on hold, something's wrong with their tools or training.
Quick Verdict
If I had to pick four metrics to watch obsessively, they'd be: answer rate (the operational foundation—nothing else matters if calls aren't getting picked up), billable percentage (the quality filter—tells you if traffic is worth having), EPC/RPC (the revenue reality—what you're actually earning per call), and conversion rate (the outcome metric—because connected calls that don't close don't pay rent).
Everything else is diagnostic. Important when something breaks, but not the metrics you need on a daily dashboard. And look—I know operators who track 40 metrics religiously and operators who watch three. The three-metric people usually make more money. They know what actually moves the needle.
For cross-channel attribution that connects calls to site behavior, JustAnalytics tracks sessions without cookie consent friction. And if click fraud is inflating your cost per call, ClickzProtect catches invalid clicks before they hit your budget.
Frequently Asked Questions
What's the difference between RPC and RPM in pay-per-call?
RPC (Revenue Per Call) measures the average payout you receive for each call sent to a buyer—total revenue divided by total calls. RPM (Revenue Per Minute) measures revenue earned per minute of connected call time. RPC is simpler and works for fixed-payout campaigns. RPM matters when payouts scale with call duration, like legal intake where longer calls signal qualification. Neither is better—they answer different questions. Track both.
How do I calculate billable percentage?
Billable percentage equals billable calls divided by total calls, times 100. If 200 calls arrive and 156 meet your buyer's duration threshold (say, 90 seconds minimum), your billable percentage is 78%. The remaining 22% were too short to generate payout. Low billable percentage usually means traffic quality issues—callers hanging up before qualification—or a duration threshold that's too aggressive for your vertical.
What's a good answer rate for pay-per-call campaigns?
Depends on vertical and buyer capacity, but 85-95% is healthy for most contact centers. Below 80% signals staffing issues, routing problems, or hours mismatch. Emergency services (plumbing, HVAC) typically run higher because callers are motivated. Considered purchases (legal, insurance) see more voicemails and callbacks. Track answer rate by hour—most operators discover their 2-4pm slump is killing 10% of daily calls.
What's the difference between connect rate and conversion rate?
Connect rate measures whether a caller reaches a live agent—calls answered divided by calls attempted. Conversion rate measures whether that connected call achieves a business outcome—appointments booked, policies sold, retainers signed. You can have 95% connect rate and 8% conversion rate if your agents aren't closing. Connect rate is operational; conversion rate is revenue. Optimize connect rate first (it's a prerequisite), then work on conversion.
Try VeloCalls for Your Vertical
AI calling + 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. Per-minute pricing — Managed starts at 4¢/min, BYOC at 2¢/min, both drop as you scale.