cloudbase.foundation — post-launch roadmap

Date: 2026-08-07 · Owner: Jonathan, with board input where marked · Status: Draft for review — Ursula, Jacob, James Continues: the launch spec (CBF-site/docs/.../2026-08-03-go-live-critical-path-design.md) at "Track C"

The site launches to a deliberately narrow bar: no lies, no dead ends. Anything broken or misleading got fixed; anything merely thin shipped thin. This is the work that follows. Every item is something we removed on purpose, something the legacy site does that the new one doesn't yet, or a known gap we are carrying knowingly.

Not covered here: the hero image and remaining copy, which are on today's critical path.

State this morning

Checked against the running services on 2026-08-07, not from notes.

Effort: S ≈ under a day, M ≈ a few days, L ≈ a week or more. Content items are usually gated on someone writing something, not on engineering time.

Tier 1 — Close the two-site overlap

Two live sites with different content and a split search presence. An accepted trade — redirecting now would delete public access to content that exists nowhere else — but the window should end on purpose rather than drift.

1.1 Migrate Past Projects · L · blocks 1.2. thecloudbasefoundation.org/current-projects-3 holds the history since 2010. The biggest content gap and the only reason the legacy site can't be switched off. Needs the project list off Squarespace, photos where they exist, and a call on how much detail each archived entry carries.

1.2 Redirect the legacy domain · M · depends on 1.1. Permanent 301 with a path-level map so inbound links and search results land sensibly:

1.3 Decide on Videos, Documents, private pages · decision. Three legacy sections with no equivalent here: rebuild, fold in, or let the redirect drop them. Videos may be worth keeping — the community actually engages with it.

1.4 Retire Squarespace · S · depends on 1.2. Ends a recurring cost. Worth confirming the renewal date so we don't pay for another term by accident.

Tier 2 — Restore what we dropped for launch

2.1 Newsletter signup → ListMonk · M. The box was removed because it silently discarded every address. The legacy site has a working signup, so this is a straight regression, and it sits on one of the four visitor actions the site exists to support. The design is done and the dangerous failure mode — double opt-in creating subscriber rows while the confirmation email fails to send — was found and the SMTP prerequisite verified on production. Only the build remains.

2.2 News bodies, then article pages · M · content-gated. All three posts have real headlines and excerpts but bodies reading "Full post content to be added" — which is why the cards have no "Read more" link. Write the bodies first, then build /news/[slug]. The engineering is small; the writing is the work.

Tier 3 — Credibility and completeness

Nothing here is broken. These are where a donor or grantmaker looking closely finds the site thinner than the organization.

3.1 Board bios and photos · S, content-gated. Six members listed with names and titles and nothing else. For a 501(c)(3) asking strangers for money, the people behind it are part of the case. The page already handles empty bios cleanly, so this is purely collecting material. Suggest 3–4 sentences each — who they are, their connection to free flight, what they bring — plus a headshot. Consistency matters more than polish.

3.2 Publish the governance documents · S · ⚠️ read before publishing. The /documents/ page is built but lists nothing. First file is the IRS determination letter, which grantmakers and corporate matching programs ask for routinely. Care required: IRS correspondence — both the CP 575 notice and the Letter 947 determination — routinely carries a responsible officer's home address, in our case a founding board member's. Publish a redacted copy, never the original.

3.3 Spanish news posts · S · content-gated. Spanish is a committed locale with full page-level parity, but the three posts are English only. The gap compounds with every new post.

3.4 Native-speaker review of Spanish copy · S · Ursula. Standing policy, not yet done for the Privacy Policy and Terms — written and translated in-house under launch pressure, and the most legally sensitive Spanish text on the site.

3.5 Legal review of Privacy and Terms · decision. Both describe only what the site actually does — no analytics or trackers, donations handed to GiveLively, submissions stored in our own CRM. Accurate, but no lawyer has read them.

Tier 4 — Data and maintenance debt

4.1 Source impact stats from Twenty · M. Country and project counts are hard-coded in four files and kept in sync by hand. This has already put one wrong figure on the live site.

4.2 Restore /resources/ when the groundcrew app ships · S. A trigger, not a bug — the page and content are intact behind a redirect. Noted so it isn't rediscovered later as a mystery.

4.3 Correct PRODUCT.md · S. Still describes the board as placeholders, the contact email as empty, the socials as dead, and an X account as real — all fixed or disproved at launch. Stale internal docs cause bad decisions later.

4.4 Board data consistency · S. Two of six entries are missing the bio and photo fields rather than carrying empty ones. Harmless now, a trap when bios land. Fold into 3.1.

4.5 Reconcile the focus-area taxonomy · decision. The Theory of Change and the site/CRM use different lists — the ToC adds Technology and says "Community Development" where the site says "Community." A governance question about what CBF says it funds, not a technical one.

4.6 Decide an accessibility standard · decision. Currently recorded as explicitly undecided, so no unearned conformance claim is made. A meaningful share of the open-call audience reaches us on slow mobile connections in remote regions, which makes this more than compliance. WCAG 2.1 AA is the usual nonprofit bar.

Decisions for the board

1. Does US work count toward the country claim? The site says 7 countries, including the active US project (Chelan) — defensible, since a visitor can count seven flags. If it shouldn't count, the number becomes 6 and that project moves to Past Projects. The two change together, never just the number.
2. Videos, Documents, private pages — rebuild, fold in, or drop? (1.3)
3. Legal review of Privacy and Terms — commission it, or accept as-is? (3.5)
4. Board bios — who writes them, what length, are we collecting headshots? (3.1)
5. Focus areas — site changes to match the ToC, or the reverse? (4.5)
6. Accessibility — commit to a standard, or keep it open? (4.6)

Proposed sequence

Newsletter leads: it's the only item that is both a live regression and fully specified. Past Projects runs alongside because it's the long pole and the rest of Tier 1 queues behind it.

Tier assignments are a proposal, not a decision.