Mantara

Inside the engagement

A business that is mostly offline

Mantara sells luxury door hardware — handles, knobs and hooks — through a busy showroom and a growing online store, with every order flowing through one Shopify instance. Roughly three quarters of revenue is offline. That structure is exactly what makes loose attribution dangerous: when the platforms claim credit, there is a real order system to check them against.

So the engagement started not with campaigns but with a read: both ad accounts and the store, side by side, going back thirty-eight months.

Reading the platforms against the store

Across 2026, platform-reported revenue ran at roughly twice what the store actually booked — month after month. That does not make the platform numbers useless; it makes them a ranking signal rather than a revenue line. We built a baseline that reconciles every month of ad spend against Shopify revenue, and set the brand's breakeven return from its real product margins instead of a platform benchmark.

That one change reframed every decision that followed: a campaign is not 'working' because it claims a return — it is working when the store's number says so.

Rebuilding the accounts

The Meta account was restructured by product line — handles first, then knobs and hooks — under one naming convention, so anyone reading the account can see what each campaign is for. The September 2026 audit showed why concentration matters: the hero handles line was booking orders at a cost roughly 40% below the account average.

On Google, the same audit surfaced a 3× efficiency gap between two campaigns of the same type running side by side, and a conversion pool polluted by low-value store-visit and chat events that had been teaching the bidding the wrong lesson. Budget moved to the efficient campaign; the conversion actions were flagged for cleanup before any automated bidding gets trusted.

The photoshoot the brand already owned

Luxury hardware does not sell as a packshot — it sells as a door, a room, a detail someone wants in their own house. Instead of commissioning a new shoot, we curated the brand's existing library in a digital asset manager: lifestyle-first images, de-duplicated, organised by product line, each cut in 1:1 and 9:16.

From those curated boards we built a 39-ad catalogue campaign covering the scale-up SKUs — every ad drawing on photography the brand already owned. The shoot strategy was editing, not shooting.

What runs now

A monthly tracker reconciles platform spend with store revenue down to the SKU. The founders and the agency read the same sheet. Spend scales on lines the store confirms, and platform claims stay in their own clearly-labelled column.