Digital TransformationUser Experience & Accessibility
By How a fixed-fee findability study takes one agency website from an org-chart navigation and an unknown page count to a content inventory, card sorts and tree tests with real users, and a validated sitemap, taxonomy and redirect map the build team can implement.
Most agency websites were not designed so much as accumulated. Every office added its pages, every reorganization renamed a few sections, and every relaunch moved the content into a new template without asking whether anyone could find it. The result is familiar: tens of thousands of pages, navigation that mirrors the org chart, three pages that say three different things, and a contact center answering questions the website already answers.
The usual response is to fund a redesign or a CMS migration. Neither fixes findability on its own. A migration moves content; it does not reorganize it. A new visual design over the same structure produces the same lost users in a better-looking site. The 21st Century Integrated Digital Experience Act (2018) asks agencies for websites that are searchable, consistent, not duplicative and designed around user needs. That is a structural requirement, and structure can be tested before anyone builds it.
This paper is for the web or digital director who champions the site, the OCIO digital services or web operations lead who owns the platform, the program manager or contact-center owner who feels the cost, and the COR who buys the work. It describes Avalon Technology Services' Information Architecture service, which starts with one engagement:
The illustration later in this paper works through the effort. The base case is nine weeks and 486 hours at borrowed market-norm rates, or $57,240 to $91,260, with a $74.3K planning fee at the midpoint, well under the $350,000 simplified acquisition threshold.
Avalon studies and recommends. The study builds nothing, migrates nothing and certifies nothing. Decisions about the structure, retiring content and acceptance of the work remain the agency's.
Any two of these phrases in one meeting usually mean the site has a structural problem nobody has tested.
| What the agency says | What it usually means |
|---|---|
| "People can't find anything on our site." | The navigation is organized around the agency, not the tasks users came to do |
| "We have tens of thousands of pages and nobody knows what's current." | No inventory, no owners, and no rule for retiring content |
| "Our search is useless." | Search is compensating for a structure that does not work, and is indexing duplicates |
| "Every office wants its own link on the homepage." | Labels and placement are settled politically, without user data |
| "The call center keeps answering questions the website already answers." | The answer exists but sits where users do not look, under a label they do not use |
| "Our navigation basically mirrors our org chart." | Users must understand the agency before they can find a form |
Three routing questions keep this service distinct from its neighbors. If the problem is the screens of one application, that is Avalon's separate UI/UX Design or User Testing & Feedback Analysis service. If it is Section 508 conformance, that is Accessibility Modernization. If it is a service journey across phone, mail and web, that is CX Strategy Development. This study restructures the content and navigation of one website.
| Authority | What it says | What it means for this study |
|---|---|---|
| 21st Century IDEA Act (2018) | Agency websites should be searchable, consistent, not duplicative and designed around user needs and data | The study produces tested, documented progress against those requirements for one site |
| U.S. Web Design System and digital.gov guidance | Shared navigation patterns and guidance for federal websites | The structure uses USWDS navigation, breadcrumb and page patterns where the site uses them |
| Section 508 (2017 refresh incorporates WCAG 2.0 Level AA) | Criteria such as 2.4.5 Multiple Ways, 2.4.6 Headings and Labels and 3.2.3 Consistent Navigation depend on structure | The structure sets the preconditions for those criteria; it is not a conformance audit |
| Plain Writing Act (2010) | Public content in clear, plain language | Navigation labels are written in users' words and tested |
| Paperwork Reduction Act (44 U.S.C. ch. 35) | OMB clearance for collecting information from 10 or more members of the public; agencies often use generic clearances | A dependency: the research pathway is decided in week one by the agency's PRA officer |
| Federal records rules (36 CFR ch. XII) | Records are disposed of only under an approved schedule | Every page flagged for removal goes to the records officer first |
WCAG 2.1 and 2.2 are current W3C guidance; the federal legal standard remains the one the 2017 refresh incorporates. One note on what this work is not: information architecture is not part of the Risk Management Framework, maps to no security control and contributes nothing to an ATO. The one security benefit is hygiene. The inventory surfaces orphaned pages and forgotten subdomains, which the agency's security team may want to know about.
The study covers one named public website or portal and one primary audience. It works from the agency's own analytics, search logs and CMS exports.
| Workstream | Activities |
|---|---|
| Inventory and analytics | Crawl of the site, content inventory with a redundant, outdated and trivial (ROT) flag on every page, Digital Analytics Program traffic overlay, search-log review of top, failed and zero-result queries |
| Top tasks | Stakeholder interviews and a top-task exercise that ranks what users come to do |
| Card sort | One open or hybrid card sort with the primary audience, designed to fit the agency's PRA pathway |
| Structure and taxonomy | Sitemap, navigation and hub pages on USWDS patterns, URL structure, plain-language labels, and a field-level taxonomy and metadata schema mapped to the CMS |
| Tree test | Unmoderated tree test of the proposed structure against top tasks, success and directness targets agreed in advance, one revision cycle |
| Migration and governance | Page-level redirect map, removals flagged for records review, ownership matrix, review cadence and search recommendations |
The card sort shows how users group the agency's content and what they call it. The tree test then strips away visual design and asks users to find answers in the proposed structure, task by task. Before the test runs, the agency's decision-maker agrees success and directness targets for each task. After it runs, the scores are reported as found, the structure is revised once where tasks failed, and the final structure ships with its scores attached.
Card sorts typically use 15 to 30 participants per audience. With members of the public, that crosses the Paperwork Reduction Act's threshold of 10, so the study either runs under a generic clearance the agency holds or is designed with nine or fewer public participants and labels any staff proxies as proxies. That is why the scope holds to one primary audience: each added audience multiplies recruiting and clearance work.
The inventory gives every page a recommended action: keep, merge, rewrite or retire, each traced to crawl, analytics or research evidence. Retiring is a recommendation, never an action. Every page proposed for removal is listed for the records officer before anything is unpublished, and every moved page gets a 301 redirect in the migration map.
| Deliverable | Format |
|---|---|
| Content inventory and audit workbook | Spreadsheet, one row per page |
| Findings report (analytics, search logs, top tasks) | Written report |
| Card sort and tree test results | Report plus raw data (CSV) |
| Validated sitemap | Diagram plus spreadsheet |
| Navigation and labeling specification | Written specification |
| Taxonomy and metadata schema | Spreadsheet, field level, mapped to the CMS |
| URL redirect and migration map | CSV or spreadsheet |
| Governance recommendations | Ownership matrix and review cadence |
| Executive readout | Briefing deck |
A content model specification for the CMS team is an option. Every deliverable is produced to conform with Section 508, is written to be vendor neutral, and belongs to the agency.
The study stops at the specification. Visual design, components and prototypes belong to Avalon's UI/UX Design service. Usability testing of designed screens belongs to User Testing & Feedback Analysis, Section 508 audits and remediation to Accessibility Modernization, and cross-channel journeys to CX Strategy Development, all separate Avalon services. CMS implementation, templates, search-engine configuration, rewriting pages at scale, ongoing SEO work and license resale are out of scope.
The accurate statement is narrow: the study produces tested, documented progress against the IDEA Act's searchable, consistent, not-duplicative and user-data-driven requirements for one site, and the structural preconditions for the WCAG navigation criteria. It does not make the site compliant with anything.
| Accurate | Not accurate |
|---|---|
| A structure tested with real users, scores attached | Makes the website "IDEA-compliant" |
| Labels and navigation that support WCAG navigation criteria | Delivers Section 508 compliance |
| Search recommendations from the agency's own logs | Search will be fixed |
| A page-level plan the build team can execute | Guarantees fewer calls or higher task completion |
| Question | Answer |
|---|---|
| Does the research need PRA clearance? | The agency's PRA officer decides in week one: a generic clearance the agency holds, or a design with nine or fewer public participants |
| What about privacy? | Card sorts and tree tests collect minimal personal information; nothing is recorded without a consent process, and data stays in agency-approved storage |
| Is Section 508 testing included? | No; conformance testing is a separate service |
| Will content be deleted? | No; removals are recommendations, flagged to the records officer first |
| Is there a conflict of interest? | Deliverables are vendor neutral, and Avalon accepts organizational conflict of interest mitigation language for any later build |
| How is it accepted? | Per deliverable, against criteria in the SOW, including the tree-test targets agreed in advance |
| Phase | Timing | What happens |
|---|---|---|
| Inventory and analytics | Weeks 1–2 | Crawl, inventory, analytics and search-log access, PRA pathway decided, dependencies dated |
| Research | 2–3 weeks | Stakeholder interviews, top-task exercise, card sort |
| Structure and taxonomy | 2–3 weeks | Sitemap, navigation, labels, taxonomy and metadata schema |
| Validation | 1–2 weeks | Tree test, one revision cycle, final scores |
| Handoff | 1 week | Redirect map, governance recommendations, executive readout |
The total is seven to eleven weeks. Recruitment and the PRA pathway drive the research phase, so the clock is tied to a dated dependency schedule. A slip moves the schedule, not the price.
| Included | Separate work |
|---|---|
| One public website, one primary audience | Added audiences, sites or languages, priced as change orders |
| Navigation and labeling specification | Visual design, prototypes and building the site |
| Redirect and migration map | Running the CMS migration or rewriting pages |
| Structural preconditions for WCAG navigation criteria | Section 508 conformance testing |
| Public, unclassified content | Intranets (priced separately) and classified environments |
Avalon takes one engagement at a time, works business hours, and works on unclassified properties only. Avalon's founder manages the engagement and the dependency schedule. A named IA/UX lead, hired or subcontracted for the engagement, with a portfolio of card sorts and tree tests, leads the research and the structure, supported by a subcontracted UX researcher. Both are identified before the proposal and listed in it. Avalon has no past federal engagements under this service. FAR 15.305 says an offeror without a record of past performance may not be evaluated favorably or unfavorably on that factor; what protects the agency is named personnel, a fixed price, a small scope and deliverables it can inspect.
| Requirement | How it is handled |
|---|---|
| IA/UX lead and researcher | Named personnel whose experience and investigation level match the agency's requirements, listed in the proposal |
| Research tools | A card sort and tree test platform the agency holds or approves, subscribed per engagement as a pass-through cost; nothing is resold |
| Access to analytics and the CMS | Accounts the agency provisions; work stays in the agency's own tools |
| Participants | Recruited with the agency's help or approval; Avalon holds no research panel |
| Insurance and contract vehicles | Any coverage the solicitation requires is a condition of award; Avalon does not hold a GSA Schedule or a governmentwide contract today |
Key personnel substitution requires CO approval. Avalon never evaluates a structure it later builds.
| Engagement | Market-norm range | Avalon |
|---|---|---|
| Findability study: one website, one audience, card sort and tree test, 7–11 weeks | $60K–$150K | Firm fixed price, milestone billed; illustrative $74.3K |
| Findability assessment: inventory, analytics and recommendations, no primary research | $20K–$45K | A smaller first step |
| IA as a line item inside a redesign or migration order | $15K–$50K | Priced at the same rates |
Large estates and multi-site consolidations run higher and are not the entry engagement. The study sits under the $350,000 simplified acquisition threshold and can be bought as an open-market purchase. IA inside a larger redesign or migration order can be subcontracted to Avalon by a prime. Avalon does not hold a GSA Schedule or a governmentwide contract today, and it does not resell software or licenses.
The best time to buy is before the migration contract is scoped, so the build team prices a tested structure instead of a copy of the old one. A study with non-severable deliverables also suits a July to September purchase from operations and maintenance funds.
The figures below are a planning illustration for one representative unclassified public website with one primary audience. They are not a quote and not a result from any engagement.
The IA/UX lead works at three-quarters time, the UX researcher at half time and the project manager at one-tenth time, at 40 hours a week. The IA knowledge base lists no rates, so each role borrows a band from the related UX services: $120 to $190 an hour for the lead, $110 to $185 for the researcher and $140 to $185 for project management.
| Role | Rate band | Base hours |
|---|---|---|
| IA/UX lead | $120–$190/hr | 270 |
| UX researcher | $110–$185/hr | 180 |
| Project manager | $140–$185/hr | 36 |
| Total | 486 |
| Scope | Hours | Rate-band range | Midpoint | Tool cost |
|---|---|---|---|---|
| Compact (7 weeks) | 378 | $44.5K–$71.0K | $57.8K | $480 |
| Base (9 weeks) | 486 | $57.2K–$91.3K | $74.3K | $630 |
| Extended (11 weeks) | 594 | $70.0K–$111.5K | $90.8K | $750 |
The base case is 486 hours, or $57,240 to $91,260, with a $74.3K planning fee at midpoint rates, inside the $60,000 to $150,000 band. The compact case runs $44,520 to $70,980; at low rates it falls below the band floor, which is a reminder that the band is a market norm, not a formula. The extended case runs $69,960 to $111,540. Tool subscriptions at an assumed $300 a month add a few hundred dollars.
As a cross-check, the knowledge base's minimum loading (half time for the lead, a quarter for the researcher) over seven weeks gives 238 hours, or $28,420 to $44,730. That lands in the $20,000 to $45,000 assessment band, not the full study, which is why the assessment drops the primary research.
The inventory changes what a migration costs. The page counts and ROT shares below are assumptions that crawl data replaces in the first two weeks; content-owner review assumes 2 minutes a page.
| Site | Pages crawled | ROT share | Pages to migrate | Records review rows | Owner review hours |
|---|---|---|---|---|---|
| Small site | 5,000 | 20% | 4,000 | 1,000 | 167 |
| Mid-size site | 15,000 | 35% | 9,750 | 5,250 | 500 |
| Large site | 40,000 | 50% | 20,000 | 20,000 | 1,333 |
A migration scoped before triage prices every page. One scoped after triage prices only what survives, with records review as a dated dependency rather than a surprise. No findability outcome is claimed here; this is page arithmetic.
The model prices the study; it claims no saving. The agency can test the fee against its own contact-center cost. At an illustrative $8, $15 or $25 per assisted contact, the $74.3K fee equals 9,288, 4,953 or 2,972 contacts, about 774, 413 or 248 a month over a year. Whether that many shift to self-service is something the agency measures, not something Avalon promises.
| Pitfall | How to avoid it |
|---|---|
| "While you're at it, rewrite the pages" | Content production is a priced option, not part of the study |
| Design or build creep | The deliverables table goes into the SOW verbatim; the study stops at the specification |
| Political navigation | One decision-maker; tree-test data settles disputes |
| Card sort without a tree test | The structure is not validated until users find things in it |
| Removing content without records review | Every removal flagged to the records officer first |
| Agency dependencies slipping | A dated dependency schedule; slips move the clock, not the price |
| Compliance oversell | No claim that the site becomes IDEA- or 508-compliant, or that search is fixed |
Most hard-to-use agency websites are not missing a redesign. They are missing a structure that was tested with the people who use it, before anyone paid to build it.
The natural first step is the fixed-fee findability study: one website, one audience, seven to eleven weeks, under the simplified acquisition threshold, ending in a validated sitemap with its test scores, a taxonomy the CMS team can implement and a page-level migration map. Where primary research is not yet possible, the findability assessment is a smaller start. To discuss a site, contact Avalon Technology Services at contact@AvalonTechServices.com.
Card sort
A research method in which participants group content items and name the groups, showing how users think about the content.
Tree test
A research method in which participants find answers in a text-only version of a proposed structure, measuring success and directness per task.
Directness
The share of tree-test participants who reach the right answer without backtracking.
ROT
Redundant, outdated or trivial content flagged during the inventory.
Taxonomy
The controlled set of categories and metadata fields used to organize and tag content.
Top task
One of the few tasks most users come to a website to complete.
All values are planning assumptions for one representative unclassified public website. None is drawn from a specific agency, a real site or a completed engagement.
| Assumption | Value | Basis |
|---|---|---|
| Property | One public website, one primary audience | Planning assumption |
| Duration | 7, 9 or 11 weeks at 40 hours a week | KB phase durations |
| Loading | Lead 0.75, researcher 0.5, PM 0.1 of full time | KB team ranges; full loading |
| Rates | Lead $120–$190, researcher $110–$185, PM $140–$185 an hour | Borrowed from related UX services |
| Tool cost | $300 a month, pass-through | Assumed |
| Triage inputs | Page counts, ROT shares, 2 minutes a page | Assumed; crawl data replaces them |
| Price check | Base fee $74.3K | Inside $60K–$150K; under the SAT |
