Paid Advertising

23% YoY revenue growth and 25% decrease in CAC

Milligram had momentum and nothing holding it up. Both paid accounts were being rebuilt every few weeks around a promo and launch calendar. Over ten months we moved Meta and Google off the calendar and onto a sustainable evergreen structure, and split the catalogue so every category became its own investable line. We also moved reporting onto real online revenue instead of in-platform numbers, so budget calls were made on what the business actually earned. Revenue grew 23% year on year off a sustainable restructure, while the cost of acquiring a new customer came in 25% cheaper.

The Milligram storefront in a shopping centre: a wide stone-clad entrance with the Milligram name and a ruler motif above, timber shelving of stationery either side and staff at a central display inside

+23%

YoY revenue growth

-25%

New customer CAC

+26%

Orders, half new customers

+160%

Evergreen ad spend scaled

The challenge

Rebuilt every few weeks around the next promo

Milligram is an Australian premium stationery and design retailer selling online and through physical stores. The brand was already well established, with real momentum and seven figures in online revenue. What was missing was structure.

Both the Meta and Google accounts ran to a promotional and launch calendar instead of a foundation the platforms could keep learning from. On Meta, 71.3% of spend sat in dated, time-boxed campaigns for seasonality and launches, with little going to evergreen. Every time a promo ended, signal was lost, learning reset, and the next campaign started cold.

On Google, one catch-all All Products campaign took the majority of the spend. The core ranges and best sellers, the biggest revenue driver in the catalogue, were buried inside it with no category-level visibility at all. Every dollar of return aggregated into one number.

What was the problem

No persistent foundation the platforms could learn from. Every campaign was time-boxed, so the algorithms never had a base that outlasted the promo it was built for.
No category-level visibility on where return came from. One All Products campaign meant one blended ROAS, and no way to see which ranges were actually earning the spend.
No dedicated brand search campaign. People already searching for Milligram were bought inside the same campaigns as cold new customers, flattering every number.
No revenue number the platforms couldn’t inflate. Budget decisions were made on in-platform reporting rather than on what the business actually earned online.

The solution

We stopped rebuilding and started compounding

Instead of a new campaign for every promo or launch cycle, we built a sustainable evergreen strategy on the right foundations, then used the promos as a secondary driver. We split the catalogue so every category became its own line with its own target, which let budget follow margin instead of a marketing calendar. Brand search was isolated so the category campaigns stayed a clean new-customer layer, and reporting moved onto real online revenue so budget calls were made on what the business actually earned.

The whole approach came down to one principle: let the accounts compound instead of starting over.

Inside a Milligram store: a long table of open books and magazines in the foreground, white shelves of colourful stationery along the wall and the Milligram name lit above a welcome message

What we built

Each piece was built to compound rather than reset the data signals, so learning carried across the whole ten months instead of starting again.

Meta flipped to 60% evergreen

From 71% of spend in promo campaigns to a 60% evergreen base, with audience quality compounding over time. Promos and launches kept, but as bursts on the base. Additive, not the engine.

Google split by category

The catch-all All Products campaign replaced with purpose-built campaigns, the core range split out as the primary revenue engine and every category given its own target.

Brand search isolated

Brand terms given their own campaign and negated out of every category campaign, keeping the core and category campaigns as a clean new-customer acquisition layer.

Measurement the business could use

MER and new customer CAC became the monthly decision metrics, with platform numbers as the tuning signals. No more peaks and troughs from stopping and starting.

The result

Revenue up 23% with barely any extra budget

Ten months into the new build, the accounts stopped resetting and started compounding. Revenue grew 23% year on year on only 13% more spend, orders were up 26%, and the cost of acquiring a new customer fell around 25%. Almost half of all orders came from first-time buyers, so the growth came from new demand rather than squeezing the existing base.

It worked because persistent structure beats reactive promos. The strongest gains landed in the months furthest from the rebuild, which is what compounding looks like, and Milligram now budgets by category ROAS instead of by whatever is next on the calendar.