Causal voice-of-customer

Know what to fix.
And exactly what it's worth.

Ranked fixes in $/year with confidence bands your team can defend. Reviews and survey in; causal estimates priced under assumptions you control.

Input
Reviews + survey
Method
Causal DML
Output
$/yr + band

$10M rev · +1pp sat ≈ +0.2% rev · assumptions set by your team

Product / experience quality

+11.0pp · 7% neg

~$221K

$210K to $232K

Service responsiveness

+9.3pp · 25% neg

~$185K

$174K to $197K

Digital experience

+9.0pp · 12% neg

~$179K

$167K to $191K

Environment & facilities

+8.9pp · 38% neg

~$178K

$166K to $189K

Environment is loudest in reviews; quality pays more per point fixed. Share ≠ worth.

Illustrative output

Example teardown

$10M rev · +1pp sat ≈ +0.2% rev · assumptions set by your team

Ranked fixes by causal worth per year
FixLiftWorth / yrat your model
Quality
7% neg
+11.0pp
~$221K
$210K to $232K
Service
25% neg
+9.3pp
~$185K
$174K to $197K
Digital
12% neg
+9.0pp
~$179K
$167K to $191K
Environment
38% neg
+8.9pp
~$178K
$166K to $189K
Support
29% neg
+7.7pp
~$155K
$141K to $169K
Onboarding
24% neg
+6.8pp
~$137K
$124K to $149K

Ready to scope a pilot for your unit?

Request early access

Pilot snapshot

What an early teardown surfaces

Anonymized pattern from a multi-unit service brand pilot. Illustrative structure only — your ranked list is priced under assumptions your finance team sets.

Surprise

Environment ranked loudest in reviews but quality had the highest $/year per point fixed.

Decision

CX reordered Q3 initiatives using causal worth, not negative mention share.

Outcome

Finance signed off on one pilot fix with a documented confidence band and assumption sheet.

Composite anonymized pilot notes. Not a client endorsement.

Method

What runs under the teardown

01

Ingest

Reviews + driver-satisfaction survey for one brand unit. Validated CSV upload.

02

Tag & corroborate

Themes in open text, checked against survey items. Both must agree.

03

Estimate causally

Double machine learning estimates satisfaction lift if each issue is resolved.

04

Price under your model

You set revenue and how a satisfaction point maps to money.

05

Rank to fund

Ordered fix list with $/year and confidence band, ready to forward internally.

Typical VoC

“Theme X appears in 12% of negative feedback.” Volume does not say whether fixing X moves satisfaction or rides along with something else.

EvidenceOS

“Resolving service responsiveness lifts satisfaction +9.3pp, worth ~$185K/yr under your model, band $174K to $197K.”

Early access program

Onboarding multi-unit service brands for paid causal teardowns. First pilots include assumption workshops and deck-ready output for internal buy-in.

  • CX / ops / finance buyers
  • One unit → pilot
  • Illustrative demo on site
Early access

One unit to start

Request a teardown

Three steps: request access here, send one unit's data when we reply, receive ranked fixes in $/year with confidence bands.

  1. 01

    Request (this form)

    Who you are, company, pilot context.

  2. 02

    Send data (after we reply)

    Reviews + satisfaction survey for one brand unit.

  3. 03

    Get teardown

    Ranked fixes in $/year under your assumptions.

Step 1 here: tell us who you are. We'll send review + survey requirements after reply.

Pilot, ongoing, and enterprise scopes. Pricing set with early customers.