Blended & multi-channel forecasting
Slides done? Here’s the same idea in a bit more depth — the part worth keeping.
In depth: four channels, four planning problems
The instinct when contacts arrive across calls, chat, email and messaging is to total them up and forecast one big number. It feels tidier, and it’s wrong, because a “contact” means something completely different on each channel. They arrive in different shapes, they’re handled with different concurrency, and some must be answered now while others can wait. Blend them and you don’t simplify the problem — you erase the distinctions that decide how many people you need and when.
The three things that don’t transfer
First, arrival shape: calls peak mid-morning, email piles up overnight and lands first thing, chat tracks web traffic and runs late into the evening. Four channels, potentially four different intraday profiles. Second, concurrency: an agent handles one call at a time but two or three chats at once, so chat volume simply doesn’t convert into headcount the way calls do — ignore it and you’ll massively over-staff chat. Third, urgency: a call is a real-time problem you staff to a service level with queueing maths; email and async are deferrable, so you staff them to a backlog and a deadline using throughput logic. Same operation, genuinely different mathematics.
Forecast apart, then blend only where staff really flex
The right sequence is to forecast volume, AHT and profile per channel, convert each into a workload or requirement using the model that suits it, and only then combine — at the requirement level. And the size of the blended number depends entirely on one honest question: can the same person actually switch between channels? Universal agents pool capacity and need fewer people overall; dedicated teams don’t. The answer is about how flexible your workforce genuinely is on the day, not how flexible the org chart claims it is.
The principle to remember: forecast each channel on its own terms, convert with the right tool, and blend only where staff truly flex. Adding raw contacts together is the fast road to a plan that’s wrong in both directions at once.
Quick quiz
Five questions. Pick an answer to each, then check your score.
1. Why can’t you just total all channels and forecast one number?
Different shapes, concurrency and urgency — adding them blurs every distinction that matters.
2. What does concurrency mean for chat?
Concurrency means chat doesn’t map to headcount like calls — ignore it and you over-staff chat badly.
3. How do real-time and deferrable channels differ in staffing?
Different planning problems — Erlang/service-level for live channels, throughput against a deadline for deferrable ones.
4. What’s the right order of operations?
Model each channel apart, convert with the right tool, and only blend at the requirement level.
5. What does the blended requirement depend on most?
Universal agents pool capacity; dedicated teams don’t. Real flexibility — not the org chart — drives the blend.
Size live-channel staffing with the Erlang C calculator, or go deeper in The Forecasting Masterclass.