Larkfield University Department: SEO Recovery and Content Governance
Larkfield University's Department of Environmental Science runs a 340-page site on a legacy CMS. Six faculty members have edit rights to their own sections; three update regularly, two have not logged in for a year, and one pastes content from Word, producing inline styles and duplicated markup. A crawl last month found 22% broken internal links, the same graduate-program description live at three URLs (/grad, /prospective/grad, /node/1187), and course lists published only as scanned PDFs. A year ago the site moved to a new CMS and every URL changed to a /node/#### pattern; organic search traffic has since fallen 30%. The university now mandates WCAG 2.1 AA conformance and a brand refresh. The department has no dedicated web staff and a modest budget for a contractor.
1. Diagnose the 30% organic decline and lay out a remediation plan, ordered by expected impact.
Read the worked answer
Diagnosis: the migration replaced descriptive, keyword-bearing URLs with /node/#### identifiers and, evidently, did not map the old paths to the new ones. Every inbound link, bookmark, and search result pointing at the old URLs now hits a 404. That transfers no ranking signal and destroys the pages' accumulated authority - a textbook self-inflicted decline. The duplicate graduate-program description adds a second problem: three URLs compete for one query and dilute each other.
Ordered plan:
- Rebuild the old URL inventory from analytics, server logs, the archived sitemap, and top referring pages, then map each old URL to its new equivalent and implement server-level 301 redirects. This is the highest-impact step and it gets worse the longer it waits.
- Fix internal links at the template level so navigation and department listings point to the new paths directly rather than through redirects.
- Resolve the duplicate graduate content: choose one canonical URL, 301 the other two to it, and update internal links.
- Submit a fresh XML sitemap and check robots.txt and any staging noindex tags that survived the migration.
- Monitor coverage and query reports weekly; recovery typically takes weeks to months, not days.
Expectation setting: 301s pass most - not all - link equity. Some rankings will not return, so plan new content for the gaps. Do not stack a URL restructure on top of the brand refresh in the same release; sequence them.
2. Design a content governance model for six faculty editors with wildly different skills and no dedicated web staff.
Read the worked answer
The root cause is that untrained editors have unrestricted control over presentation. Governance should change what they are allowed to control, not just how often they are trained.
Structural fixes:
- Replace free-form pages with structured content types - faculty profile, course, news item, publication - each with fixed fields. Structured content cannot carry pasted inline styles because there is nowhere to put them.
- Restrict roles in the CMS: faculty get author or contributor rights with no access to templates, raw HTML, or global navigation. A single coordinator publishes.
- Strip the paste-from-Word path by enforcing plain-text paste or providing a clean-paste editor.
Human fixes:
- A one-page style guide covering headings, dates, file naming, and image sizes, plus a 20-minute onboarding clinic.
- A named owner per section, with a quarterly review to flag stale pages - the two dormant accounts should be demoted or removed.
- The contractor builds and maintains templates and handles migration cleanup; faculty own facts, not layouts.
Trade-off to state explicitly: faculty lose visual freedom and gain reliability. Sell it as protection - their content stops breaking when templates change, and the department stops being embarrassed by dead links. Track link error rate and stale-page count as the two governance metrics.
3. The course lists exist only as scanned PDFs. What is wrong with that, and what would you do about it given the accessibility mandate and a mostly mobile audience?
Read the worked answer
What is wrong: a scanned PDF is an image of text. It has no text layer, so screen readers announce nothing, browsers cannot reflow it, text cannot be resized or recolored, and search engines cannot index course names. It fails WCAG requirements for non-text content and for information and relationships, and it is the single worst format for a phone user who must pinch and pan around a fixed page.
The fix, in order:
- Publish course lists as structured HTML - a real table with header cells, or a definition list - generated from the CMS course content type. This makes it searchable, responsive, and accessible by default.
- If a downloadable PDF is required by policy, generate it from the same structured data as a tagged, text-based PDF with a correct reading order and document language, not a scan.
- If any legacy scans must stay temporarily, run OCR, tag them, and provide an accessible HTML alternative next to the link, with a clear label on the file type and size.
- Never let the PDF be the only path to the information.
Governance tie-in: the moment course data lives in structured fields, the accessibility problem stays fixed permanently, because every future PDF is regenerated rather than re-scanned. That is the argument to make to faculty who prefer uploading documents.