Multi-skill routing and blended agents
Multi-skill routing, where agents handle contacts from more than one queue, is a common efficiency lever that makes staffing significantly more complex. This guide covers the tower model vs. blended routing trade-off, the correct Erlang C approach for multi-skilled agents, and the occupancy and AHT effects you must account for.
Tower model vs. blended routing
Most contact centres use some combination of dedicated and blended agents. Understanding the trade-off is essential before building a staffing model.
Tower model
Dedicated agent pools per skill
- •Each skill has its own agent pool
- •Agents handle only one skill type
- •Erlang C run independently per skill
- •SL managed per queue separately
- •Simple intraday management
Advantage
Simple to plan, schedule, and manage. Agent performance is easier to measure.
Disadvantage
More total headcount, because idle agents in one skill cannot absorb overflow from another. Higher FTE requirement for the same total volume.
Blended routing
Agents handle multiple skill queues
- •Agents assigned to 2+ skill queues
- •ACD routes to any available eligible agent
- •Higher occupancy, less idle time
- •Typically 5–15% fewer total agents needed
- •More complex intraday management
Advantage
Lower total headcount for the same volume. Idle capacity in one skill absorbs peaks in another.
Disadvantage
AHT creep on secondary skills. Higher occupancy risk if blending pushes combined load above cap. More complex to schedule and monitor.
The correct staffing model for multi-skilled agents
The most common mistake in multi-skill staffing is blending volumes and AHTs from different skill groups into a single Erlang C model. This produces the wrong answer for every skill simultaneously.
Correct approach: per-skill Erlang C
Run Erlang C per skill group
Use each skill's own volume, AHT, and SL target. Get seated agents required per skill.
Apply secondary-skill AHT uplift
Agents handling a secondary skill are typically 15–30% slower than the team average. Adjust the AHT input for skills where many agents handle it as secondary.
Sum per-skill requirements
Total seated agents = sum of all per-skill requirements. This is slightly conservative but accurate for planning.
Apply shrinkage once
Divide total seated agents by (1 − shrinkage). Do not apply shrinkage per skill and then sum; this double-counts.
Check combined occupancy
Verify that the blended occupancy across all skills (total Erlangs ÷ total seats) does not exceed your occupancy cap (typically 85%).
AHT creep on secondary skills
When agents handle a skill they are less practised in, their AHT for that skill is higher than the team average. This is AHT creep: it reduces the effective capacity gained from blending and must be accounted for in the staffing model.
Why AHT creep happens
Agents navigate systems more slowly on secondary skills. Resolution paths for secondary skills are less familiar, so there is more hold time while the agent looks up the answer. Knowledge base use is higher. The effect is strongest in the first 3–6 months after skill certification.
How to measure it
Pull per-agent, per-skill AHT from your ACD reports. Compare the AHT for agents handling a skill as primary vs. secondary. The gap is your AHT uplift. Typical values: 10–15% uplift after 6 months on secondary skill; 20–35% in the first 3 months.
How to account for it in staffing
When running Erlang C for a skill, use a weighted AHT: (primary agents × primary AHT + secondary agents × secondary AHT) ÷ total agents. This gives a realistic effective AHT for the Erlang C calculation rather than the team average.
When to narrow the skill blend
If secondary-skill AHT creep exceeds 25%, consider narrowing the skill blend for new agents until they are 6+ months tenured on the primary skill. The efficiency gain from blending disappears if the AHT penalty offsets it.
Occupancy effects with blended routing
Blending increases agent occupancy because agents fill idle time from one queue with contacts from another. This reduces total headcount, but it also pushes occupancy higher, which increases agent fatigue and burnout risk.
Example: tower vs. blended occupancy
Inputs
- Skill A: 100 calls/hr, 5 min AHT
- Skill B: 50 calls/hr, 10 min AHT
- SL target: 80/20
Tower model
- Skill A: 10 agents, 83% occ
- Skill B: 10 agents, 83% occ
- Total: 20 agents
Blended (approx.)
- Blended pool: ~18 agents
- Combined occ: ~91%
- 2 fewer agents, higher burn risk
Blended routing saves 2 agents but pushes occupancy above the 85% guardrail. To maintain safe occupancy, you need ~19 agents (1 saved vs. tower, 88% occupancy). The efficiency gain is real but smaller than it first appears, and it must be bought within the occupancy cap.
Intraday management in a blended operation
Skill prioritisation rules
When both queues have contacts waiting, the routing engine needs a priority rule. Common approaches: highest SL deficit gets priority; longest waiting contact gets priority; primary skill always gets priority. Define and test your priority rule before go-live, because incorrect prioritisation causes one queue to collapse while the other looks healthy.
Real-time skill adjustment
Intraday WFM for blended operations may require adjusting skill assignments mid-shift. If skill A is struggling and skill B is over-staffed, temporarily reassigning skill A as secondary skill for skill B agents is a common lever. This requires WFM to have real-time visibility of both queues simultaneously, not just per-skill views.
SL reporting per skill
Blended routing makes SL reporting more complex: the ACD must attribute each contact to the correct skill for SL tracking, even if the handling agent is multi-skilled. Ensure your ACD is configured to report SL by skill group, not just by agent or queue.
Agent performance attribution
When an agent handles contacts from two skills, AHT and quality scores should be tracked per skill, not blended. An agent with a 6-minute AHT on billing and an 11-minute AHT on technical support should be coached on technical separately from billing; blended AHT reporting hides the skill-specific gap.
Multi-skill routing questions
How do you calculate staffing for multi-skilled agents?
Run a separate Erlang C model for each skill group using that skill's volume, AHT, and SL target. Sum the per-skill agent requirements. Apply shrinkage once to the total. This is slightly conservative (assumes no sharing between skills) but is the correct planning approach. Do not blend AHTs across skills; this gives the wrong result for all queues simultaneously.
What is the difference between a tower model and blended routing?
A tower model runs separate agent pools per skill: each pool handles only one skill, staffed independently. Blended routing assigns agents to multiple skills; the routing engine directs contacts to any available eligible agent. Tower models are simpler to plan. Blended routing reduces total headcount by 5–15% but increases occupancy, adds AHT creep on secondary skills, and complicates intraday management.
Does skill-based routing affect occupancy?
Yes. Blended routing increases average occupancy because agents fill idle time from one queue with contacts from another. This can push occupancy above your target guardrail. When modelling blended operations, check that combined occupancy across all skills does not exceed your maximum occupancy cap (typically 85%), not just the occupancy per individual skill.
What is AHT creep in multi-skill operations?
AHT creep is the observed phenomenon where agents handle secondary-skill contacts more slowly than their primary skill. An agent primary in billing may handle billing at 4 minutes but technical support at 8 minutes (vs. the team average of 6 minutes). Apply a 15–30% secondary-skill AHT uplift in your staffing model to account for this.
Model multi-skill staffing in Turnella
Add multiple workstreams with separate AHTs and SL targets. Each gets its own Erlang C model. See the combined headcount and occupancy in one view.
Related guides
Erlang C explained
The queue model behind per-skill staffing
Occupancy explained
The burnout risk when blending pushes occupancy up
AHT guide
AHT creep mechanics and measurement
Multi-channel calculator
Per-channel Erlang C with combined view
Capacity planning guide
Full 7-step WFM cycle
Intraday management
Real-time skill-queue management