Answer Economics — the measurement doctrine
Deflection told you what didn't happen. Cost per answer tells you what everything costs.
One trunk, many branches — a portfolio of channels
Every support organization already runs a portfolio of answer channels — specialists, self-service, community, bots, in-app help, and soon your customers' own AI agents. Answer economics is the discipline of pricing that portfolio: one blended number, and the strategy that moves volume toward the cheap, governed end of it. Deflection stays on the books — as a legacy view.
Six channels, one portfolio
Issue-centric accounting: an answer is an answer, whoever — or whatever — delivers it. Each channel has a unit cost and a prerequisite. The strategy writes itself once you see the spread.
| Channel | Unit cost / answer | What it needs to work |
|---|---|---|
| Specialist assist (1:1) | $12–30 | Staffing, and the queue that comes with it |
| Self-service | $0.30–1.00 | Findable, current, trusted articles |
| Community | $1–3 | Moderation and a validation path into the record |
| Chatbot / AI answers | $0.40–1.50 | A governed, machine-ready corpus — or it answers confidently and wrong |
| In-app / predictive | $0.10–0.50 | Context signals plus the same governed corpus |
| Customers' AI agents (API) | pennies | Stable identifiers, explicit applicability, honest refusals |
Unit costs are illustrative placeholders — the Measurement Workbook computes yours from your own volumes and loaded costs.
The blended number
Two companion ratios tell you whether the number is moving for the right reasons. The shift-left ratio: the share of answers delivered without a human in the loop. The known-vs-new mix: how many questions were already covered by the record versus genuinely new — because the goal is not fewer questions, it's that known questions stop reaching specialists at all.
The inversion your dashboard will show
When shift-left works, the easy questions leave the assisted queue first. What remains for your specialists is harder by construction — so assisted time-to-resolve rises, assisted complexity rises, and the cost of an assisted answer rises. By design.
Leaders who aren't told this in advance read success as failure and kill the program at exactly the moment it's working. Put the inversion in the executive brief before the first dashboard ships.
Where deflection fits
Deflection counted the tickets that didn't happen — a negative space, argued about in every budget meeting it ever entered. It isn't wrong; it's just not the successor. Deflection · legacy view
If you budget by deflection today, don't fight the religion: run both views for two quarters. The successor metric wins the argument on its own, because it prices the channels you're investing in instead of estimating the ones you avoided.
Price your own portfolio.
The Measurement Workbook — 16 tabs of cost-per-answer economics with the channel inventory, scenario modeler, and the metric-inversion brief — ships in the Contextkeeping Toolkit ($79 alone, $299 for everything). Or have the baseline built for you from your real exports: it's one of the six deliverables in AnswerEcon.