Performance ads, SEO and checkout work that returned 3–4x on ad spend and lifted average order value by 10–15% for one of Singapore’s best-known homegrown ice cream brands.

Ice cream is the definition of an impulse buy, and almost none of the things that make it easy to sell in a shop survive the move online. In a scoop store the decision takes ten seconds and the product is in someone’s hand; on a website the same person has to choose flavours they cannot taste, commit to a delivery window, and trust that a frozen product will arrive frozen. That last part quietly governs the whole category. Frozen delivery carries a real cost per drop, and it does not care whether the basket holds one tub or six — which means a single-pint order can cost more to fulfil than it earns, and a channel can look busy while making no money at all. On top of that, demand is weather-led and occasion-led in a way that resists planning: a hot weekend, a birthday, a late-night craving, a party that needs an ice cream cake. Meanwhile the delivery aggregators own the impulse moment outright, sitting between the brand and the customer and taking a margin for it. The task was not to prove that people in Singapore want ice cream. It was to make the brand’s own store a channel worth spending on — one where the media pays for itself and the basket is big enough to be worth delivering.

Turn online ordering into a genuinely profitable channel rather than a convenience the brand subsidises. Three measures defined it. First, a return on ad spend high enough to keep re-investing month after month with confidence. Second, a higher share of the people who reached the store completing an order instead of stalling over flavour choice, delivery slots or the cost of getting it there. Third — and the one that decides whether frozen delivery works at all — a bigger average order: more pints per basket, cakes and party formats attached, so each drop carries enough value to justify the trip. Revenue return, order conversion and average order value were the scoreboard, and they were treated as a single problem rather than three, because in this category a bigger basket is what makes a good ROAS achievable in the first place.

We built performance campaigns across Meta and Google Shopping and Search, structured around the way the range is actually bought rather than the way it is catalogued. Search carried the deliberate, high-intent moment — someone looking for an ice cream cake for a birthday, or for ice cream delivery in their part of the island — where the customer has already decided and simply needs to find the brand rather than an aggregator’s listing. Social carried the discovery and prompting work, putting flavours, cakes and party formats in front of people whose craving or occasion had not yet turned into a search, and leaning into the moments the category actually spikes on. Campaigns were weighted towards the multi-tub bundles, cakes and party packs rather than single pints, because those are the baskets that make frozen delivery viable; the feed and campaign structure were maintained so spend concentrated on the lines that converted at a workable basket size instead of being spread evenly across the menu. A retargeting layer ran constantly against browsers and cart abandoners, which matters here because an abandoned ice cream cart is very often a delivery-window question rather than a change of mind.

Alongside the paid work we developed SEO for the branded, flavour and delivery searches the brand should own outright, so a steady share of that high-intent demand arrived with no media cost attached — the part of the mix that keeps the blended return healthy when the auction gets expensive. On the store itself we reworked the ordering and checkout experience, tightening product pages and flavour selection, making delivery dates, slots and the cold-chain promise unambiguous before checkout rather than a surprise at the end, and clearing the friction in cart and payment where orders were leaking. Bundle and add-on paths were built into that flow, so building a bigger basket was the easy route through the store rather than something the customer had to work at — the single change most directly responsible for the movement in average order value. Email capture ran through both the site and the campaign journeys, and lifecycle email and retargeting flows carried those contacts back for the next occasion, so the media budget was buying a customer rather than a one-off order.

Over five to six months the campaigns returned 3–4x on ad spend, order conversion settled between 2.5% and 3%, and average order value rose 10–15% — an online channel that earns back several times what it costs, on baskets large enough to make frozen delivery worth running.

See the rest of our client work on the case studies page.