Chapter 54: Buttons at 19:17
The buy button was ready before the page carrying my name was fast enough to deserve it.
At 18:56 on Friday, both trial routes waited behind PageTurn's scheduled release gate.
The common opening time was 19:00 GMT. Paid impressions and the seventy-two-hour clock were armed for that second.
North Quay's twenty-four shop pages had passed content and payment checks that afternoon.
Then the final performance probe loaded Saltmere 1.8 seconds slower than Larkspur.
The difference appeared across mobile connections in three test regions. Desktop results alone would have concealed the defect.
One point eight seconds could decide whether an impatient customer reached a payment button.
Opening anyway would turn a temporary defect into seventy-two hours of disputed data. Later statistical adjustment could not recreate abandoned customers.
My coordinator wanted Larkspur held while Saltmere opened on schedule to compensate.
That would reverse the disadvantage rather than remove it.
PageTurn proposed delaying both routes until the page bundles matched the performance tolerance.
The opposing project had no role in creating my heavier page.
Its team would still lose the same seventeen minutes of Friday evening.
I accepted the common delay and its revised close time. Oliver and Pippa received the same notice without seeing my asset file.
The platform stamped BOTH ROUTES HELD — PERFORMANCE COMPARABILITY.
No customer could purchase through either measured page during the hold. Shop staff saw a neutral delayed-opening message rather than a blamed project.
Pre-existing preview interest remained outside conversion as yesterday's rule required.
The engineer traced my delay to an uncompressed author photograph.
Its image had been approved for appearance, not delivery weight.
She replaced the file with a visually identical optimised version.
The asset checksum changed, so the platform recorded the technical substitution before testing. The original heavy image remained in the diagnostic archive.
No title, cover, price, copy or author claim changed with it.
Saltmere's median load time fell inside the common tolerance. The improvement persisted across three consecutive probe rounds.
PageTurn repeated the probe across the same regions and device classes.
Larkspur remained closed while my page proved stable.
At 19:14, the performance report turned green for both routes.
The release controller scheduled one new auditable opening rather than clicking pages manually.
NEW OPENING — 19:17 GMT.
NEW CLOSE — MONDAY 19:17 GMT.
The seventeen lost minutes were not restored through extra exposure later. Friday's planned promotional posts were rescheduled to the shared start.
Both projects received the same seventy-two-hour duration from the new start.
At 19:17, the two purchase buttons changed from WAITING to BUY simultaneously.
PageTurn's log recorded the event to the millisecond. Neither shop route could claim a manual early click.
North Quay's shop pages received the same opening token and consumer-reach rules.
I placed no test purchase of my own.
The first settled Saltmere order appeared at 19:19 through an anonymised payment row.
By 20:00, gross orders placed Saltmere ahead in the common summary.
The public screenshot my coordinator prepared displayed only that rising total.
I asked her to leave it unpublished until the net fields matured. The draft stayed in our private queue with its time captured.
At 20:18, the first cancellation removed one paid intention from the provisional count.
Two more followed before 21:00.
The gross bar continued rising while the net line stepped backwards underneath it.
Larkspur's gross count remained lower in the same approved summary.
Its cancellation field was also still provisional and revealed no customer identities.
PageTurn stored performance probes, delay approval, asset substitution and synchronized opening log.
The payment gateway retained order, cancellation and settlement status separately. A provisional author dashboard could not overwrite those underlying states.
At 23:40, Saltmere still led first-night gross orders.
That was a useful signal and an incomplete one.
My cancellation rate had begun climbing quickly enough to narrow every celebratory screenshot.
The first night gave me the prettier number and the more dangerous subtraction.