How did cloud ACD break on-premise WFM between 2011 and 2016?
Between 2011 and 2016 the automatic call distributor moved to the cloud while the workforce management platform stayed on premise. Cloud routing on one side, on premise workforce management on the other, and middleware connecting them was the defining technical challenge of the era. It produced a new kind of gap. Not a migration gap, but an integration maintenance gap, where the operation did not rebuild the methodology because no single platform changed, and the methodology slowly drifted from what the floor was actually doing.
The drift was hard to see because it did not appear in the routing statistics. It appeared in adherence reports that no longer matched the intraday schedule. The practitioner who could diagnose a hybrid integration gap was rarer than one who could run either platform independently, because the gap lived in the connection between two systems that no implementation team had fully documented before leaving.
The same window also introduced a forecasting break. As chat, email, and short message channels entered Ontario financial services and telecom centres, the single channel Erlang C model broke at scale for the first time. Most operations deployed the channel without deploying a practitioner who could rebuild the forecast for concurrent handling. That mechanism, a new channel breaking an assumption in the existing model, would repeat exactly when AI deflection arrived years later.
Frequently asked
Why did cloud ACD break on-premise WFM?
Because the routing moved to the cloud while workforce management stayed on premise, connected by middleware that was rarely documented. The methodology drifted from the floor without any single platform formally changing.
Where did the hybrid gap show up?
Not in routing statistics, but in adherence reports that no longer matched the intraday schedule. Diagnosing it required understanding the undocumented connection between the two systems.
What broke Erlang C in this era?
The addition of chat, email, and short message channels. Erlang C was built for single channel voice, and concurrent multi channel handling required a fundamentally different forecasting approach.
Get new articles delivered
Practitioner-written WFM intelligence. No spam, no vendor pitches.