Peak-season commerce, re-platformed and re-priced

Verdant's storefront degraded badly under festive load, and the two previous peak seasons had both included a multi-hour outage during the highest-revenue window of the year.

Acquisition cost had risen for six consecutive quarters while the marketing team optimised to platform-reported return that finance could not reconcile to the ledger.

Product, engineering and the media agency were operating on three different definitions of a conversion.

A headless rebuild on Next.js with an edge-cached catalogue, static product pages regenerated on inventory change, and a queue-backed order pipeline that degrades gracefully instead of failing.

Load testing to twelve times the previous peak became a release gate. Nothing shipped without passing it.

In parallel our growth practice rebuilt measurement on server-side tracking with a single conversion definition agreed by finance, then ran a structured incrementality test across paid channels.

Roughly a third of paid spend turned out to be buying customers who were already arriving organically. That budget was reallocated to retention and to two channels that tested genuinely incremental.

The festive season ran without an incident, absorbing peak traffic twelve times higher than the previous year's failure point.

Blended acquisition cost fell 38% within nine months while revenue grew, because the spend that was buying nothing had been identified and removed.

Conversion rate rose 31%, with the largest single contribution coming from page performance rather than any change to the funnel.