Free Workforce Planning Tool
The Erlang C formula is the industry standard for determining how many agents a call center needs to meet its service-level agreement. Our free Erlang calculator lets you model staffing across daily, weekly, and monthly horizons — so you can stop guessing and start planning with data.
This free Erlang C calculator helps contact center managers and workforce planners determine optimal staffing levels. Enter your call volume, average handle time, and service level target to calculate the number of agents needed. Supports daily, weekly, and monthly planning horizons. Built by Convershake, the Voice AI platform for digital lenders, to help operations teams transition from manual staffing to intelligent automation.
Whether you manage 20 seats or 2,000, this tool applies the same Erlang C methodology used by workforce-management platforms worldwide. Pair it with Convershake's Voice AI to automate a share of those calls entirely.
Contact centers are the backbone of customer experience — and staffing them efficiently is a multi-billion-dollar challenge. Here's the latest data:
$5.4 B
Global WFM software market in 2025
Report Prime
77 %
of centers still plan inefficiently
Peopleware WFM Report 2025
12 %
annual growth in contact center AI spend
CMSWire 2026
80/20
most common SLA: 80 % of calls in 20 s
Industry standard
According to CMSWire's 2026 call center report, AI-assisted workforce management is becoming table stakes. Yet the foundation remains the same: accurate traffic modelling with the Erlang C formula. Overstaffing by even one agent per shift can cost $35,000–$50,000 per year; understaffing degrades customer satisfaction and increases churn.
The Erlang C formula calculates the probability that a caller must wait based on three inputs:
Call volume
The number of incoming contacts during a given interval.
Average handle time (AHT)
How long each call takes on average, including after-call work.
Number of agents
Available staff during that interval.
From these inputs, the calculator derives the minimum number of agents required to achieve your target service level (e.g., 80 % of calls answered within 20 seconds). Learn more about the underlying math in Peopleware's step-by-step guide.
A mid-size European lender used daily Erlang modelling to right-size their 120-agent collections team. By aligning shifts to actual call arrival patterns, they eliminated 22 redundant FTEs while maintaining a 82/20 service level. Combined with Convershake's deterministic Voice AI handling routine payment reminders, the team redirected savings into complex-case specialists.
See how Voice AI complements staffing →A business process outsourcer used weekly and monthly Erlang projections to build a transparent staffing plan for an RFP response. The granular, defensible numbers helped them beat three larger competitors and secure a 500-seat contract worth €4.2 M annually.
Explore compliance features →After a product launch tripled inbound volume, a fintech firm's WFM team used the monthly Erlang calculator to plan a phased hiring ramp. The model predicted needed headcount within 3 % accuracy each month, keeping service levels above 80/20 throughout the scale-up.
Book a demo with Convershake →Not all Erlang formulas serve the same purpose. Choosing the right one depends on your contact center's queuing behaviour.
| Formula | Assumes | Best For |
|---|---|---|
| Erlang C | Callers wait indefinitely (no abandonment) | General call center staffing — this calculator |
| Erlang B | Blocked callers are lost (no queue) | Trunk line / circuit capacity planning |
| Erlang A | Callers abandon after a patience threshold | More realistic queuing — accounts for hang-ups |
For a deep dive into all three models, see Soon's comprehensive comparison or Intradiem's white paper on Erlang-O.
Getting a number out of an Erlang calculator is easy. Getting a useful number requires careful preparation. Follow this workflow to ensure your staffing model reflects reality rather than wishful thinking.
Step 1: Gather Historical Data
Pull at least 8–12 weeks of call detail records (CDRs) from your ACD or telephony platform. You need three data points per interval: inbound call count, average handle time (AHT), and average speed of answer (ASA). Strip out outlier days — public holidays, system outages, marketing-campaign spikes — unless you're specifically planning for those scenarios. The cleaner your input data, the more reliable the Erlang output.
Step 2: Choose the Right Interval
Erlang C is most accurate at 15- or 30-minute intervals because call arrival rates can vary dramatically within a single hour. If your ACD only exports hourly data, consider splitting it using a proportional arrival-rate curve. Our daily calculator lets you model a full day of intervals; the weekly and monthly tabs help you see patterns across longer horizons and plan shift rotations accordingly.
Step 3: Set a Realistic Service-Level Target
The 80/20 target (80 % of calls answered within 20 seconds) is the most common benchmark, but it isn't universal. High-value financial services lines often target 90/10, while back-office support may accept 70/30. Your target directly affects the number of agents the Erlang calculator recommends — tightening by just 5 percentage points can add 10–15 % to required headcount.
Step 4: Account for Shrinkage
Shrinkage — the gap between scheduled and productive time — typically runs 25–35 % in most contact centers. It includes breaks, training, coaching sessions, meetings, absenteeism, and system downtime. If your Erlang calculation says you need 50 agents in a half-hour slot, you'll actually need to schedule 65–70 to have 50 productive agents on the phones. Never skip this step; it's the most common source of understaffing surprises.
Step 5: Iterate and Validate
Run the Erlang calculator with your data, then compare the recommended headcount against your actual historical performance. If the model consistently over- or under-predicts, adjust your AHT or shrinkage assumptions. Over time, you'll calibrate a model that's accurate to within 2–5 % — precise enough to drive shift-bid decisions and hiring plans.
Even experienced workforce managers fall into these traps. Recognising them early saves thousands in misallocated labour costs.
Using daily averages instead of interval data
A day with 1,000 calls spread evenly needs far fewer agents than one with a 400-call spike between 10–11 AM. Always model at the interval level.
Ignoring after-call work (ACW) in AHT
AHT must include talk time + hold time + ACW. Omitting the 30–90 seconds of wrap-up per call systematically under-counts required agents.
Treating weekends like weekdays
Call patterns on Mondays and Fridays often differ by 20–40 % from mid-week. Use the weekly tab to model day-of-week variation explicitly.
Forgetting multi-skill routing complexity
If agents handle multiple queues (e.g., sales + support), Erlang C applied to each queue independently will overestimate total headcount. Use pooled-queue modelling or simulation for multi-skill environments.
Setting it and forgetting it
Call patterns evolve as products change, seasons shift, and marketing campaigns launch. Re-run your Erlang model monthly at minimum, and weekly during volatile periods.
Not factoring in AI-deflected calls
If you're deploying conversational AI (like Convershake's Voice AI), subtract the calls handled end-to-end by automation from your inbound volume before running Erlang. This avoids overstaffing for volume that never reaches a human agent.
Not sure what numbers to plug in? These benchmarks from Call Centre Tools and industry surveys give you a solid starting point.
| Metric | Typical Range | Notes |
|---|---|---|
| Average Handle Time | 4–7 minutes | Financial services skews higher (6–9 min); tech support 8–12 min |
| Service Level Target | 80/20 to 90/10 | Premium lines target 90/10; cost-optimized lines may use 70/30 |
| Shrinkage | 25–35 % | Remote teams often see 28–32 %; on-site 25–30 % |
| Occupancy Target | 80–88 % | Above 88 % risks burnout; below 75 % signals overstaffing |
| Abandonment Rate | 3–8 % | Above 5 % typically triggers staffing review |
These benchmarks serve as starting points. Your optimal configuration depends on factors like customer expectations, regulatory requirements, and channel mix. In regulated industries such as digital lending, compliance obligations may dictate tighter service levels and longer handle times due to mandatory disclosures and identity verification steps.
Traditional workforce management treats every call as a human-handled event. But in 2026, that assumption is increasingly outdated. AI voice agents can resolve a significant share of routine interactions — payment reminders, balance enquiries, appointment confirmations — without ever reaching a human agent.
This changes the Erlang equation fundamentally. If voice AI deflects 30 % of inbound volume, you're running the Erlang calculator on 70 % of your original call load. For a 200-agent center, that could mean 40–60 fewer seats needed — a saving of $1.4–$2.1 million per year in fully loaded agent costs.
The key is deterministic deflection — knowing with certainty which call types the AI handles and which it escalates to humans. Probabilistic chatbots create unpredictable overflow; deterministic voice agents like Convershake follow scripted, compliant flows that make the remaining human workload predictable and Erlang-modelable.
Pro Tip
Run two Erlang models side by side: one with your current total volume, and one with AI-deflected volume subtracted. The gap between them is your automation ROI — and it makes a compelling business case for piloting Voice AI.
Is the Erlang C formula still relevant with AI and omnichannel?
Absolutely. While newer models like Erlang A and simulation-based planning add nuance, Erlang C remains the foundation of workforce management across the industry. AI changes the inputs (lower call volumes) but not the underlying math. For omnichannel, you run separate Erlang models per channel and aggregate.
How accurate is the Erlang calculator for small teams?
Erlang C works well for teams of 15+ agents per interval. Below that, the formula's assumption of random Poisson-distributed arrivals breaks down, and small fluctuations have outsized impact. For very small teams, consider simulation-based planning or add a larger shrinkage buffer.
What's the difference between the daily, weekly, and monthly tabs?
The daily tab models a single day's intervals — ideal for fine-tuning shift start/end times. The weekly tab shows day-of-week patterns so you can adjust Monday versus Friday staffing. The monthly tab captures seasonal trends and helps with hiring and capacity planning over longer horizons.
Can I use this Erlang calculator for chat and email staffing?
The Erlang C formula is designed for single-contact, real-time channels like phone calls. For chat (where agents handle multiple concurrent sessions) and email (which is asynchronous), you'll need modified models. However, the underlying principles — matching supply to demand — remain the same.
How often should I recalculate staffing with the Erlang formula?
Best practice is to refresh your Erlang model weekly with the latest ACD data and recalibrate shrinkage and AHT assumptions monthly. During seasonal peaks, product launches, or major campaigns, switch to daily recalculation to avoid being caught off-guard by volume spikes.
Have more questions? Check out the Convershake FAQ or Peopleware's Erlang C resource hub.
Use the Erlang calculator to find your ideal headcount — then let Convershake's Voice AI handle the routine calls your agents shouldn't be spending time on.