nanaselect

Appearance

Switching themes ​

In the admin: Themes — the screen this page documents (opens in the running app).

What/why: Presentation & theming — what this screen manages and why it exists.

Purpose: change the app's active look-and-feel — per surface, from the admin, with no rebuild or deploy.

When to use ​

  • Applying a new brand or seasonal look to the public site.
  • Restyling the admin back-office independently of the buyer experience.

Prerequisites ​

  • Admin access with the edit_data capability (any data_editor or admin role).

Steps ​

  1. Open Admin → Themes (/admin/themes).
  2. The Public theme card lists every theme registered for the public surface; the Admin theme card lists admin-surface themes. The active theme in each card carries an "Active" badge.
  3. Pick a theme with its radio button and click Apply.
  4. The change is stored in D1 (app_settings) and takes effect on the next page load — no rebuild, no redeploy. Each apply is recorded in the audit trail (entity theme, entity id public/admin).

Behavior guarantees ​

  • The two surfaces are independent: changing the public theme never restyles the admin, and vice versa.
  • Only themes registered for a surface can be selected or persisted; a stale or invalid stored value falls back to the surface default (nanawall public, admin admin) instead of breaking rendering.
  • Deleting a theme file whose slug is still stored is safe — resolution falls back to the surface default on the next request.

To add a new theme to the pickers, see Theming reference.

A/B-testing a presentation (theme + layout) ​

Instead of applying one theme/layout to everyone, you can split visitors between two or more presentation variants and promote the one that converts best — no rebuild or deploy. This reuses the same A/B experiment surface as persona-path (track) experiments; a presentation experiment just splits on the theme + layout a visitor sees rather than on their track.

Steps ​

  1. Open Admin → Experiments (/admin/experiments) and click New experiment.
  2. Under What to test, choose Presentation (theme & layout). (The subject is fixed once the experiment is created.)
  3. Add at least two variants. Each variant is a theme + layout combination; mark one as the control (the baseline the others are measured against). Each theme/layout combination may appear only once. Traffic is split evenly.
  4. Set the goal metric and the guardrails (minimum sample size per variant and minimum runtime), then Create draft experiment.
  5. On the detail screen, Start the experiment. Only one experiment (of any kind) may run at a time, so stop a running one first.

While it runs ​

  • Every genuine-human visitor is assigned a sticky variant — the same visitor always sees the same theme/layout across sessions — and the public surface serves that variant's theme + layout (overriding the admin default above).
  • The assignment respects the visitor's consent tier: it is keyed on the recognized cross-visit id where marketing consent allows, otherwise the functional session cookie. No variant dimension is recorded for a visitor below the measurement tier.
  • The Results table shows each variant's reach and goal-metric rate, the lift vs control, and the two-proportion significance test — the same readout as a track experiment, keyed per variant.

Promoting the winner ​

  • Once the guardrails are met and a variant wins, click Promote on that arm.
  • Promotion writes the winning variant's theme and layout as the public defaults (the same app_settings the Themes page sets), so once the experiment stops running the whole audience is served the winning presentation — with no code change.

Behavior guarantees ​

  • A presentation experiment never changes a visitor's track attribution — it only changes what they see. Track experiments and presentation experiments are mutually exclusive (one running experiment at a time).
  • Only themes/layouts registered for the public surface can be chosen as arms or promoted; a variant whose theme/layout was later removed cannot be promoted.
  • If no presentation experiment is running, the public surface serves the admin default theme/layout exactly as before.