Telltale RFP signal: “Modern UX” + a deliverables checklist of wireframes, comps, and a style guide. That’s not UX—that’s a beauty contest with 508 risk bolted on.
Here’s what COs/CORs actually care about: fewer failed transactions, lower call volume, clean 508 acceptance, and a transition that doesn’t torch the ATO or the CMS.
For COs/CORs: write UX that de-risks delivery
- Put outcomes in the PWS: target task success rate, completion time, error rate, dropout on key flows. Make them acceptance criteria, not nice-to-haves.
- Specify constraints up front: PRA strategy for moderated research, access to analytics/call center logs, legacy CMS limits, and USWDS adoption expectations.
- Require 508 proof, not promises: Trusted Tester process, AT coverage (JAWS/NVDA/VoiceOver), remediated sample with before/after defects.
- Downselect with a scenario exercise: give a task, 48 hours, ask for a research plan, a test protocol under PRA constraints, and a risk log. Score against L/M—not slide polish.
For small businesses: show UX that survives federal gravity
- Map roles to risks: research ops, content, IA, service design, 508 SME, analytics. One “UX unicorn” isn’t believable on a multi-year BPA.
- Baseline, then move the needle: include current metrics (DAP/GA4, call drivers), your test plan, and the delta you will hit by quarter.
- USWDS ≠ lipstick: show component-level adoption with governance (tokens, theming, contribution rules) and how you’ll migrate existing patterns.
- PRA isn’t a blocker if you plan it: propose generic clearances, intercepts, or internal user recruiting; note timelines in the sprint plan.
- Q&A to strip incumbent bias: ask to clarify ownership of design artifacts (Figma libraries, research, components), to allow remote moderated testing, to define acceptance metrics, and to open analytics/call logs early. If the SOW says “use incumbent’s proprietary design system,” ask for a government-owned alternative.
De-risk the cutover
- Content freeze and migration plan with inventories and redirects.
- Parallel run/beta period with explicit exit criteria.
- 508 backlog remediation plan with cadence and defect SLAs.
- Knowledge transfer: repos, design tokens, research repository, and RACI for approvals.
Read the history before you price
- Past TO mods and CRs hint at hidden UX debt (content, IA, 508).
- 508 audit findings show recurring defects you’ll inherit.
- Public CX dashboards/FOIA logs/call scripts reveal the real pain points.
If you could add one UX acceptance metric to L/M tomorrow, what would it be?