Telephony Capacity Planning – Optimally Dimensioning Your Resources
Too many staff during quiet periods, too few at peak times – this is the classic capacity trap in telephony. In German SMEs, an average of 23 % of personnel costs in the communications area are wasted through over- or understaffing. At the same time, during peak periods, up to 40 % of incoming calls are lost – with direct revenue consequences. This article explains how to dimension your telephony capacity precisely and where AI Voice Agents make the decisive difference.
Why conventional telephony capacity planning fails
The forecasting dilemma
The challenge of capacity planning lies in the unpredictability of call volumes. Even experienced planners make assumptions that in practice regularly deviate from actual volumes by 20–35 %. The reasons:
- Weather-related fluctuations (particularly relevant for trades, care services, delivery services)
- External events (media coverage, competitor offers, seasonal variation)
- Campaign effects from marketing activities
- Illness and holiday absences within the team
The consequence of this planning uncertainty: businesses overstaff as a safety buffer and pay for capacity they do not need – or they understaff and lose revenue.
The Erlang C formula and its limits
In professional call centres, capacity planning is calculated using the Erlang C formula. It determines the required number of staff based on:
- Call volume (calls per hour)
- Average handling time (AHT)
- Target reachability (service level: e.g. 80 % of calls answered within 20 seconds)
Example calculation:
- 40 calls/hour
- AHT: 5 minutes (300 seconds)
- Target service level: 80/20
The Erlang C formula for these values yields a required staffing level of 5 agents. However, with an eight-hour working day and 40 % variation in call volume, you would need to plan for between 3 and 7 staff simultaneously – which is simply not feasible in small businesses.
For SMEs with 5–50 staff, conventional Erlang planning is therefore often not applicable. The solution lies in a different approach.
Peak load analysis: The foundation of every capacity decision
Measuring call distribution systematically
Before making capacity decisions, you must know the actual distribution of your calls. Measure over at least four weeks:
- Hourly call volumes (when do most calls come in?)
- Daily distribution (which days of the week are peak days?)
- Weekly distribution (are there monthly or quarterly peaks?)
- Seasonal patterns (summer vs. winter, pre-Christmas, etc.)
A typical call distribution in a German SME shows the following patterns:
| Time window | Share of daily calls |
|---|---|
| 7:00–9:00 | 8 % |
| 9:00–11:00 | 28 % (main peak) |
| 11:00–13:00 | 20 % |
| 13:00–15:00 | 18 % |
| 15:00–17:00 | 22 % (second peak) |
| 17:00–19:00 | 4 % |
This distribution means: between 9 and 11 am and between 3 and 5 pm, almost half of all calls arrive. To be fully staffed during these times, you would have to employ the same staff during quieter periods – which leads to idle cost.
Recognising seasonal patterns
Depending on the industry, seasonal patterns vary considerably:
- Plumbing/heating: Peak in October/November (start of heating season), low point in July
- Tax advisory: Peak January–April (tax returns), quiet in summer
- Medical practices: Peak September–March (cold and flu season), relatively constant
- E-commerce: Peak October–December (Christmas business), often +150–300 % versus normal operation
Planning for seasonal peaks with human staff means: hiring for the peak months or using agency workers – both costly and often suboptimal in quality.
AI Voice Agents: Instant scaling without staffing costs
Scaling in seconds rather than weeks
The fundamental difference between human staff and AI Voice Agents in capacity planning: AI scales instantaneously. While hiring and onboarding a new member of staff requires 4–12 weeks, an AI Voice Agent can scale from one to a hundred simultaneous conversations within seconds.
For capacity planning this means:
- No overstaffing risk: The AI Voice Agent is only active when conversations are incoming
- No understaffing risk: Every call is answered immediately, regardless of call volume
- No overtime costs: 24/7 operation without premiums, night shift allowances, or weekend surcharges
- No sick days: 365 days a year, 24 hours a day, 100 % available
Formulae for hybrid capacity planning
In a hybrid model (AI Voice Agent + human staff), calculate the required human capacity as follows:
Formula 1: Required human capacity Human capacity = Total calls × (1 – AI automation rate) / Avg. handling time
Example:
- 400 calls/week
- AI automation rate: 65 % (routine enquiries)
- Remaining complex calls: 140/week
- Avg. handling time (human): 12 minutes
- Required staff capacity: 140 × 12 / 60 / 40 = 0.7 FTE
This means: where previously 2 full-time staff were needed for telephony, now a part-time employee suffices – plus the AI Voice Agent for routine enquiries.
Formula 2: Cost comparison Annual saving = (Previous personnel costs) – (New proportional personnel costs + Annual AI costs)
Example:
- Previous personnel costs: 2 × €45,000 = €90,000/year
- New personnel costs: 0.7 × €45,000 = €31,500/year
- Annual AI costs: €4,800 (€400/month)
- Net saving: €53,700/year (60 % reduction)
Cost comparison: Human vs. AI vs. Hybrid
| Model | Monthly cost | Availability | Scalability |
|---|---|---|---|
| 2 full-time staff | €7,500 | Mon–Fri 8–6 pm | Minimal |
| 1 part-time staff | €2,500 | Mon–Fri 9–5 pm | Minimal |
| AI Voice Agent only | €299–799 | 24/7/365 | Unlimited |
| Hybrid (0.5 FTE + AI) | €2,200–3,200 | 24/7/365 | High |
The hybrid model is the optimal solution for most SMEs: AI handles routine enquiries and night-time hours, while human staff process complex cases and build personal customer relationships.
24/7 without overtime costs: The hidden ROI
An often underestimated factor in cost calculations is the additional cost of evening and weekend coverage. In Germany, overtime premiums under collective agreements are typically 25–50 %, night work premiums 25–40 %, and Sunday premiums up to 150 %.
If a business wants to extend its reachability from Monday–Friday 9–5 pm to 24/7, this would cost the following with human staff:
- Hours outside standard time: approx. 128 hours/week
- At €20/hour base wage × 1.4 premium = €28/hour
- Additional cost per week: €3,584
- Additional cost per year: approx. €186,000
An AI Voice Agent available 24/7 costs for the same period: €3,600–9,600/year. The difference speaks for itself.
Implementation recommendation: Optimal capacity in four steps
Step 1: Four-week measurement of actual call volume (hourly, daily, weekly)
Step 2: Categorisation of call types by automation potential (appointment enquiries, information requests, complex advisory calls, complaints)
Step 3: Configure AI Voice Agent for automatable categories, define thresholds for handover to human staff
Step 4: Re-dimension staffing plan based on remaining human capacity requirements
Get started now
anicall.io helps you with capacity analysis and shows you how an AI Voice Agent optimally complements your telephony resources. Free initial consultation including capacity calculation for your business.