Pages are assembled from a 300-section library, and every section carries its exit rate and conversion lift. Publish risk is scored before go-live — slow mobile LCP, missing CTAs and empty SEO fields do not ship quietly.
A section with 46% exit and negative lift is not a design opinion. It is a row in a table, with sessions next to it.
pages.cedxsystems.com — live build
Runs on demo data. The storefront, sections and session figures are the demo's.
3.6M sessions in 30 daysacross 262 live pages · avg CVR 2.3%
320 pages · 300 sectionsa versioned library, not one-off pages
17 A/B experiments runningpublish risk scored before go-live
What it is
Three things a page system has to get right.
Drafts assembled from your own library
Describe the page — winter clearance landing with urgency banner — and the draft is assembled from the section library: pattern matched from the bank, SEO title, description and handle pre-filled, predicted CVR band on the row. The screen says it plainly: deterministic demo, no live model.
Draft queue: fall teaser 1.8–8.8% predicted CVR
Outlet clearance: 7-section stack, 4.1–10.3%
Push lands the draft as a Draft row in the library
pages — screen-2
300 sections, scored by what they cost
Every section carries uses, position, exit rate, CVR delta and sessions. The buy box runs on 48 pages at 14% exit and +1.6 lift. One reviews block exits 59% of visitors — and the table says so next to its 151,425 sessions.
145 sections with negative CVR lift
69 sections at drop ≥ 40
Filter by hero, video, product grid, footer
pages — screen-3
Themes are versioned releases
18 themes with versions, status, templates, sections, pages and owners. The main storefront runs 2.4.1 across 248 pages; five themes sit in preview, staged; older releases stay archived, not deleted.
7 themes live serving traffic
5 preview staged, archives kept
Templates, sections and page counts per version
pages — screen-4
Product tour
Four screens, captured from the running build.
Not a mockup and not a concept deck. This is what opens at /app/pages.
pages.cedxsystems.com
01 — Overview
Publish risk is a gate, not a suggestion
91 pages carry publish risk, 13 of them critical — SEO gaps, missing CTAs, mobile LCP over 3.5 seconds on go-live candidates. The root-caused alerts name the offenders: a scheduled fall landing with empty discovery metadata, a collection template at 3.6s mobile LCP.
Publish risk 91 · 13 critical
Section types ranked by drop score
Storefront health 70, computed on screen
02 — Generate
A draft in the queue by lunch
Three drafts in the queue — fall drop teaser, holiday gift guide, outlet clearance — each with its section stack and predicted CVR band, each pushable into the pages library with a Generated badge.
23 sections reused from the library
Predicted CVR bands, not promises
SEO fields pre-filled from the brief
03 — Sections
The library keeps score
Sorted by sessions, the table is a league of reusable parts: promo bar on 12 pages, trust-badges strip on 30, a footer that exits 55% with zero lift. Sections that hurt get found because they are measured.
Exit % and CVR Δ on every row
Position: above fold, mid, below fold, footer
248 active sections of 300
04 — Themes
Ship a look like a release
Theme versions move from preview to live with their templates, sections and pages attached — 2.4.1 on 248 pages, a locales pack on 86, a wholesale theme on 55. Rollbacks exist because archives do.
Version, status and owner per row
Live, preview and archived as filter chips
18 themes, 7 serving traffic
Who runs it
Three roles keep the library honest.
Roles, not references. We have no named customers yet, so nobody in these photographs is quoted, credited or claimed as one.
Content lead
Writes against the section scores — retires the block that exits 59% instead of debating it.
300 sections · 145 hurting
Merchandiser
Owns the templates and the theme releases, and keeps the collection pages under the LCP line the gate enforces.
18 themes · 7 live
Growth marketer
Runs the A/B queue and reads publish risk before anything is scheduled — 13 critical pages is a to-do list.
17 experiments · risk 91
The shape of it
What the demo storefront actually looks like.
Every figure below is legible in the captures above. Nothing here is a projection of your estate — it is the state of the demo data.
3.6Msessions in 30 daysavg CVR 2.3% on published pages
262live pages of 320storefront health 70
300sections in the library248 active · 145 hurting CVR
9.7%best predicted CVR bandupper bound on generated drafts
Template mix, share of the page library320 published pages by template
Landing34%
Product22%
Collection16%
Other16%
Blog12%
Pages live262 of 320 serving traffic
Live · 262
Draft or staged · 58
Storefront healthlive · CVR · publish and drop criticals
70
Health 70 — computed from the four inputs
Publish critical 13 · drop critical 6
How it runs
A page's life, in the order it actually happens.
01
Assemble
A draft is built from the section library — stack pattern, SEO fields and predicted CVR band attached.
02
Score
The publish-risk gate reads SEO gaps, missing CTAs and mobile LCP before the page can be scheduled.
03
Test
A/B experiments run against live traffic — 17 in flight on the demo storefront.
04
Refresh
Conversion-drop rankings surface the sections that leak, and the library gets edited, not the excuse.
One record
Pages reads the record the estate already keeps.
The storefront is not an island of templates. Its sections, forms, search terms and session data are the same ones the rest of the estate uses.
Finding this out on the third call is worse for you than reading it here, and worse for us.
Pages is not generally available. What opens today is the live build running on demo data.
The Generate screen is deterministic — it says no live model on itself. Nothing on this page claims AI-written copy; drafts assemble existing sections from your library.
Predicted CVR bands come from the demo's scoring model. They are planning inputs, not performance promises.
We have no named customers to show you, so this page shows none. The storefront in the captures is the demo workspace, not a reference.
Session, LCP and publish-risk figures are the demo site's data, shown so you can read the mechanics.
No audit or compliance certification has been issued for Pages. What we can evidence about hosting, encryption and access is on the security page.
Yes. Every screenshot is a capture of the running build and you can open the same build at /app/pages. It runs on demo data — the storefront in the captures is the demo one.
Does Generate write the copy?
No, and it says so on screen: deterministic demo, no live model. A brief is matched to a section-stack pattern from the library, SEO fields are pre-filled, and the draft lands in the queue with a predicted CVR band.
What is publish risk?
A score computed before go-live — SEO gaps, missing CTAs, slow mobile LCP on candidates. 91 pages carry it in the demo, 13 critical, and the alerts name the pages.
How are sections scored?
By what they cost: exit rate, CVR delta and sessions on every row of the library. 145 sections currently show negative lift, which is precisely the point of measuring them.
Is Pages audited or certified?
No certification has been issued. What we can evidence about hosting, encryption, tenant isolation and production access is written up on the security page.
The storefront is running. Go and look at it.
Live build, demo data, no card. Then ask which of your sections would show up with negative lift.