$work/ikisakiIKISAKI

A storefront that holds its speed through campaign peaks

Built against a performance budget, released behind a rollback measured in seconds.

IKISAKI — hero image
ClientIKISAKIIndustryRetail & DTC CommerceOutcomePerformanceYear2025
At a glanceCampaign traffic was exposing a storefront that had never had a performance budgetWe set one per template and enforced it in CI, then rebuilt around itReleases roll back in under a minute, so campaigns are no longer a risk event
TBCMobile LCP under load

Objective

IKISAKI's peaks are short and expensive: a campaign lands, traffic multiplies, and the storefront decides whether that spend converts.

They wanted speed that survives the peak and releases that can be undone.

Objective image A
Objective image B

Solution

A performance budget per template, agreed before design and enforced in CI — including a cap on third-party bytes with an owner recorded against every script.

Image, font and script strategy set once, properly, rather than negotiated per page.

Platform operations handled: environments, secrets, instant rollback, and a runbook the client's team holds.

Solution image A
Solution image B

Outcome

TBCMobile LCP, real usersTBCRollback timeTBCThird-party scripts removed

The storefront holds its budget under campaign load, and a bad release is a minute-long problem rather than an afternoon.

Four third-party scripts were removed during the audit, which was the single largest speed win.

What’s next

Experimentation on the collection page, and edge personalisation by market.

Deleting four scripts did more than six months of tuning.IKISAKIEcommerce manager
Quote or film still