Does live chat actually remove voice contact?
The deflection assumption inside every chat business case
Every chat business case rests on the same assumption: the chat traffic replaces voice traffic, so the operation can run with less voice capacity. The measured reality across UK operations that have scaled chat is more complicated. Some chat traffic is genuine substitution from voice. Some is incremental traffic that wouldn’t have happened on voice. Some is customer behaviour shifting in ways that don’t cleanly add or subtract. The honest answer to “does chat remove voice?” is: partially, and the proportion depends on how you implement it.
What the measurement actually shows
Operations that have measured chat-launch effects carefully report a recurring pattern. For every 100 chat contacts handled, the breakdown is roughly:
40–55 are genuine substitution from voice. Customers who would have rung; they chose chat instead. Real deflection, real voice-cost reduction.
25–40 are incremental contacts. Customers who wouldn’t have rung at all — the friction of voice put them off, or the question wasn’t urgent enough. Chat’s lower friction surfaces contacts that voice never saw. Not deflection; net new demand.
10–20 are channel-shift only. Customers who would have used self-service but ended up on chat instead. Real substitution but from a cheaper channel; net cost-up.
The conditions that change the proportion
Five factors materially shift the substitution rate.
1. Where chat is offered. Chat embedded only on the contact-us page substitutes more from voice. Chat surfaced proactively on product pages, FAQs, or checkout produces more incremental volume.
2. The customer demographic. Younger digital-confident customers substitute more readily; older customers either don’t adopt chat or end up using it incrementally for things they’d have lived without otherwise.
3. Contact-reason mix. Transactional contacts substitute heavily; complex contacts barely substitute at all (customers still prefer voice for complex).
4. Self-service quality. Strong self-service deflects simple voice contacts to FAQ. Chat then pulls some of those self-service customers up the cost curve.
5. Wait-time differential. If chat is staffed for low wait while voice has long queues, substitution is high. If both have similar waits, substitution drops materially.
The honest capacity-planning approach
Plan capacity on three assumptions, not one.
Voice volume reduction. Assume 40–55% of chat volume is voice substitution. Voice volume falls by that proportion of chat volume. Plan voice capacity accordingly — not on the assumption that voice and chat are perfectly fungible.
Total demand uplift. Assume total contact volume rises by 25–40% of chat volume (the incremental contacts). Total cost goes up unless chat’s lower per-contact cost more than offsets the volume rise.
Cost per contact. Chat per-contact cost is lower than voice but not by as much as the marketing suggests — concurrency drag, longer total resolution time on complex chat, QA cost. Real chat per-contact cost is usually 60–75% of voice, not 25–40%.
Conclusion
Chat removes voice contact partially — usually about half of chat volume is genuine voice substitution. The other half is incremental volume and channel-shift from cheaper channels. Capacity plans built on the “100% substitution” assumption under-staff voice and over-promise cost saving. The honest plan models all three flows: substitution, incremental, channel-shift. Operations that do this scale chat profitably; operations that don’t are surprised by their voice queues.
Pair with pros and cons of chat, staffing for chat, capacity planning when mix is changing, and gen-AI for planners.