Win One Diet, Then the Next: How Food is Good Scaled Horizontally on Meta

By Kevin Veitia

Client

Food is Good (FIG), an app that scans food and checks it against your diet, allergies and nutrition goals

Channels

Meta Ads, with Google App campaigns tested

Timeframe

November 2025 to early 2026

Engagement

Meta Ads audit, web-to-app quiz funnels and paid acquisition

Goal

Cost per free-trial start, by diet

The challenge

Food is Good, or FIG, is an app for people who have to read every label. Scan a product in the store and it tells you whether it fits your diet, your allergies or your nutrition goals: gluten-free, low FODMAP, anti-inflammatory and many more. FIG says it supports more than 2,000 dietary restrictions and has more than 50,000 users.

When we audited its Meta account in November 2025, most of its history had gone to the wrong goals. About two-thirds of the spend went to campaigns optimized for free-trial starts, but fewer than 2 in 100 of those trials became a paying subscription. About a quarter went to campaigns optimized for landing-page views, which were cheap and almost never turned into customers. Looked at by audience instead, broad campaigns and Advantage+ took about two-fifths of the budget between them, for very little in return.

One line in the audit stood out. Audiences built around a specific diet had the best cost per acquisition in the account by a wide margin, as the account recorded it: about a seventh of what the creative-testing campaigns paid, and about a fiftieth of what Advantage+ paid. The people who knew exactly what they couldn't eat were the ones who needed the app.

Double proof

An ad account can tell you what worked. It can't always tell you why. So before rebuilding anything, we went into FIG's product analytics: which diets people chose when they set up the app, and which of those users went on to convert. Celiac disease and gluten-free came out on top.

That's what we mean by double proof. The product data and the ad results pointed at the same people, from two sources that don't depend on each other. When that happens, you stop testing everything and commit. We switched off the ad sets that weren't built around a diet and put the budget behind one diet first.

What we did

Won one diet before adding the next

Each diet is its own small market, with its own worries, its own vocabulary and its own moment of need. Someone newly diagnosed with celiac disease and someone told to try low FODMAP may want the same app for completely different reasons. So we learned the celiac journey end to end first: which hooks stopped the scroll, which words people used for their own situation, and where they dropped out of the funnel. Only then did we open the next silo, low FODMAP, then anti-inflammatory, each with its own audiences and its own creative.

That's scaling horizontally: growing by opening new silos with a playbook you've already proven, instead of pushing more and more budget into the one that works until it gets expensive.

Built quiz funnels between the ad and the app

Instead of sending people straight to an app store, we sent them to a short web quiz for their diet that ended in a free-trial offer, in versions for web, Android and iOS. A quiz lets the ad make one specific promise, and lets the person see the app understands their diet before they commit to anything.

Let the message match the diagnosis

The winning ads were specific. My doctor told me to go low FODMAP brought in free trials at $28.10, the best message in the account. Ads that named a worry, like not knowing whether a food is anti-inflammatory, came in at about $28 too, and a direct if you're gluten-free, you need this app at about $30. The worst was the most generic: stop wasting time checking for gluten, at $121.63. Same diet, same app, four times the cost. Saving time is a benefit everyone claims. A doctor's instruction is a moment people actually live through.

Talking-to-camera video did most of the work. About half of all free-trial starts in the first month came from videos fronted by FIG's own growth lead, each speaking to one diet. A creator's grocery haul, walking a supermarket and scanning products, brought in 160 trials at $33.13. The same creator's haul in a different store cost more than twice as much, which is why every video gets its own test.

Trusted Android, questioned iOS

The biggest gap wasn't between ads. It was between phones. Android ad sets brought in free-trial starts for $18.62 to $29.96. The same approach on iOS cost $115.77 to $146.82, about five times more. iPhone owners aren't five times less interested in their diet. The likelier explanation was measurement: since iOS 14.5, much of what happens after an iOS ad click is hidden from Meta, so the iOS campaigns were optimizing half-blind. Until the tracking was fixed, the budget went where results could be seen.

Google had the same problem in another form. An old search campaign reported a 105% conversion rate, more conversions than clicks, which means something was being counted twice. Google App campaigns ran, but we didn't scale a channel whose results we couldn't verify.

Results

In the first month, 4 November to 3 December 2025:

• Diet-specific Android ad sets brought in free-trial starts at $18.62 (low FODMAP), $23.82 (gluten-free) and $24.94 (anti-inflammatory, the biggest, with 455 trials)
• Celiac and gluten-free ads produced 486 trial starts, and low FODMAP ads 427
• Facebook brought in trials more cheaply than Instagram, at $41.42 against $48.34, and Facebook Reels gave the best mix of volume and cost
• Threads and Audience Network were cut, at $106 and $122 a trial

One honest caveat. Every figure above is a cost per free-trial start, not per paying customer. Trial-to-paid wasn't tracked reliably while we were there, and the audit had found that fewer than 2% of trials converted. A cheap trial is only worth what it turns into, which is why the plan was to optimize for purchases as soon as tracking allowed.

Why it ended

The first month was very good and the second was decent. FIG asked us to scale, and we did. Not long after, FIG changed course on paid acquisition, and the engagement ended earlier than planned.

If we ran it again, we'd tie each step up in budget to paying customers rather than free trials, so every increase was judged on revenue. That takes purchase tracking that works, which is why it belongs at the start of an engagement, not the end. Our piece on how long ads take to work explains how we pace those decisions.

What you can take from this

• Look for a second source of proof. When your product data and your ad results point at the same customers, commit.
• Win one silo before you open the next. Each diet, audience or market needs its own playbook.
• Specific beats general. A doctor's instruction beat 'save time' four to one.
• A five-times gap between iOS and Android is usually a tracking problem, not an audience problem.
• A cheap trial is only worth what it turns into. Optimize for paying customers as soon as tracking lets you.
• Judge each step up in budget on paying customers, not on free trials.

Want a second set of eyes on your account? Book a free 30-minute consultation and we'll walk through what's working, what isn't, and where the next win is. Curious what it costs first? Our pricing is public.