Agency analytics
AccessForce Insights
See exactly where people get stuck.
Accessibility stops being an annual audit and becomes something an agency can watch, measure and improve.
Every figure on this page is illustrative prototype data or a projected model. None of it is a measured AccessForce outcome. Once an agency runs a real experience, its own values would be labelled “Measured”, and measured and illustrative figures are never mixed in the same view.
Analytics are aggregate and privacy-conscious: counts, times and drop-off by step, language and selected experience preference. No individual answers, documents or identities are exposed here.
Completion funnel
IllustrativeStarted
10,000
Completed language selection
9,850−150
Program selected
9,400−450
Identity completed
8,900−500
Income completed
7,100−1,800
Documentation completed
5,800−1,300
Submitted
5,300−500
Biggest drop-off
Income & household information
1,800 of the 10,000 people who started stop at this step — the largest single loss in the process.
Where are people struggling?
IllustrativeFriction is inferred from aggregate signals: abandonment rate, “I don't understand” clicks, help requests, answer revisions, time spent, language switching, human escalation, validation errors, repeated explanations, “I don't know” responses.
Household income
Cognitive load score 34/100
- Avg. time
- 4m 21s
- “I don't understand”
- 18%
- Escalation
- 11%
- Answer revised
- 23%
Required supporting documents
Cognitive load score 41/100
- Avg. time
- 3m 48s
- “I don't understand”
- 15%
- Escalation
- 9%
- Answer revised
- 12%
Household composition
Cognitive load score 47/100
- Avg. time
- 2m 52s
- “I don't understand”
- 12%
- Escalation
- 6%
- Answer revised
- 19%
Prior applications and claim history
Cognitive load score 55/100
- Avg. time
- 2m 20s
- “I don't understand”
- 10%
- Escalation
- 5%
- Answer revised
- 9%
Language-specific analytics
Is the experience equally good in every language?
| Language | Completion | Average time | Help requests |
|---|---|---|---|
| English | 78% | 14 min | 12% |
| Spanish | 61% | 19 min | 25% |
| Chinese | 67% | 18 min | 21% |
| Vietnamese | 58% | 21 min | 29% |
| Arabic | 60% | 20 min | 27% |
| Other | 63% | 19 min | 24% |
Accessibility disparity detected
Spanish-speaking applicants are 17 percentage points less likely to complete this process.
See how multilingual access worksAccessibility preference analytics
Which accessibility needs are still being failed?
Reported in aggregate by the experience preference someone chose, so barriers can be found without exposing anything about an individual.
Simpler language
62% completion versus 78% overall
Biggest drop-off: Income verification
The verification step may be reintroducing higher-complexity source language after the conversational experience.
Screen reader
66% completion versus 78% overall
Biggest drop-off: Document upload
Upload status and error messages may not be announced. Review live-region updates and error wording on this step.
Large text / high contrast
74% completion versus 78% overall
Biggest drop-off: Review and submit
The review summary may reflow poorly at larger text sizes. Check the summary layout at 200% text.
Voice input
69% completion versus 78% overall
Biggest drop-off: Household income
Spoken amounts are being rejected by strict formatting. Accept natural phrasing and confirm the interpreted value.
Extended time
71% completion versus 78% overall
Biggest drop-off: Documentation
People returning after a break may lose their place on this step. Confirm the saved position is restored exactly.
Keyboard navigation
75% completion versus 78% overall
Biggest drop-off: Program selection
Check focus order and visible focus on the program cards.
Before and after
Did the redesign actually work?
Completion
58%
82%
Improved
Average time
27 min
15 min
Improved
Help requests
31%
14%
Improved
Abandonment
42%
18%
Improved
AXS
43
93
Improved
Illustrative before/after values for the redesigned experience — not measured outcomes.
Accessibility heatmap
Every step, scored
Every level is written out as a word as well as shown visually, so nothing here depends on colour alone.
Discover service
AXS 71
- Completion
- 94%
- Average time
- 1m 40s
- Help requests
- Low
- Drop-off
- Low
Program selection
AXS 68
- Completion
- 92%
- Average time
- 2m 05s
- Help requests
- Medium
- Drop-off
- Low
Identity
AXS 74
- Completion
- 89%
- Average time
- 3m 10s
- Help requests
- Low
- Drop-off
- Medium
Household
AXS 52
- Completion
- 84%
- Average time
- 4m 05s
- Help requests
- Medium
- Drop-off
- Medium
Income
AXS 42
- Completion
- 71%
- Average time
- 4m 21s
- Help requests
- High
- Drop-off
- High
Documentation
AXS 47
- Completion
- 58%
- Average time
- 6m 30s
- Help requests
- High
- Drop-off
- High
Review
AXS 66
- Completion
- 55%
- Average time
- 2m 15s
- Help requests
- Medium
- Drop-off
- Low
Submit
AXS 78
- Completion
- 53%
- Average time
- 1m 05s
- Help requests
- Low
- Drop-off
- Low
AI improvement loop
AccessForce gets better by learning where the experience fails.
Step 1
Residents use AccessForce
Step 2
Aggregate friction signals
Step 3
AI identifies accessibility patterns
Step 4
Recommendations generated
Step 5
Agency approves changes
Step 6
Experience improves
Step 7
AXS reruns
Accessibility becomes measurable, observable and improvable.
Call-driver analytics
Turn support calls into product insights.
| Why people call | Calls per month | Avoidable? | Accessibility opportunity |
|---|---|---|---|
| What documents do I need? | 12,400 | Likely self-serviceable | Proactively explain required documents during the application. |
| Which form do I need? | 9,800 | Likely self-serviceable | Let people describe their situation and suggest the likely process. |
| What does this question mean? | 8,600 | Likely self-serviceable | Offer plain-language rewording and “why are you asking?” in every language. |
| Did you receive my application? | 7,200 | Likely self-serviceable | Confirm receipt clearly and show status without a phone call. |
| I need another language. | 6,400 | Likely self-serviceable | Let the resident choose or switch language at any point in the process. |
| How do I change my answer? | 4,900 | Likely self-serviceable | Allow answers to be revised at any step without restarting. |
| I can't access the PDF. | 3,700 | Likely self-serviceable | Offer the process as an accessible conversation instead of a document. |
| I don't understand eligibility. | 3,100 | Keep human support | Explain the rules in plain language — the agency still makes the determination. |
| I need a person. | 2,400 | Keep human support | Keep human support one step away, with everything already saved. |
Cost-to-serve calculator
IllustrativeEnter your own numbers. Nothing is stored, and nothing here is a savings guarantee.
- Total contact cost
- $1,875,000
- Navigation-related share
- $656,250
- Potential addressable support cost
- $956,250
Illustrative model. This is a scenario calculation using the numbers you enter — not a guaranteed savings estimate, and not a measured AccessForce result.
Traditional cost drivers
- PDF translation
- Language vendor
- Call-center staffing
- Interpretation
- Repeat contacts
- Manual navigation help
With AccessForce
- AI self-service
- Approved multilingual terminology
- Human escalation
- Analytics
- Continuous improvement
Projected agency impact
Projected Agency Impact
Multilingual support calls
35% fewer form-navigation calls
Translation maintenance
One source process instead of many PDFs
Form abandonment
24 point reduction in abandonment
Repeat questions
25% reduction in repeat contacts
Average support time
Shorter calls, already-saved context
Digital completion
40% faster digital completion
First-contact resolution
More issues resolved on first contact
Language access
Every covered language measured, not assumed
Projected model values for a prototype, never presented as measured AccessForce outcomes.