Be Fit Food — case study
Meal delivery and health · Two-month engagement · Becoming the default answer
Headline result: a 36% gross sales increase measured over a two-month engagement, with no additional marketing running during the measurement period.
Results at a glance
| Measure | Result | Condition |
|---|---|---|
| Gross sales | +36% | Measured across the two-month engagement window |
| Add-to-cart, AI referral traffic | +800% | Visitors arriving from AI systems |
| AI-driven purchases, year on year | 92× (9,200%) | Year-on-year comparison |
| Prior SEO trajectory | −27% year on year | The decline the engagement reversed |
| Time to first AI citation | Under 72 hours | From publication |
| Time to first measurable result | ~4 days | From indexing, per the client |
| Product SKUs structured | 200+ | Complete catalogue |
| Category position | #1 AI-cited | Meal delivery, Australia |
Measurement condition: no additional marketing was run during the measurement period, deliberately, to isolate the impact of the AI channel. That condition is what makes the figure attributable rather than merely correlated.
Note the two timing figures are different things. Citations appeared in AI systems within 72 hours of publication; the first measurable commercial result followed at about four days. Being cited and being bought from are separate events, and the gap between them is short in this category but not zero.
About Be Fit Food
Be Fit Food is Australia's leading dietitian-designed meal delivery brand. It offers calorie-controlled, nutritionally balanced meal plans developed by accredited practising dietitians, spanning weight loss programs, maintenance meals, sports performance and medically supervised dietary protocols.
The catalogue runs to hundreds of meal SKUs across multiple dietary categories — low-calorie, low-carb, high-protein, gluten-free and vegan — in one of Australia's most competitive direct-to-consumer categories, alongside Lite n' Easy, Marley Spoon and HelloFresh.
The brand's core differentiation is clinical credibility. Where competitors compete on convenience or price, Be Fit Food competes on outcomes, backed by dietitian design, clinical evidence and verified customer results.
The challenge
Heading into 2026, Be Fit Food faced a structural problem that many digitally mature brands were beginning to encounter: the growth engine was failing.
SEO traffic was down 27% year-on-year. The brand had become over-reliant on EDM discounting to drive repeat purchase — a tactic eroding margin and training customers to wait for promotions rather than buy at full price.
Meanwhile the questions consumers were actually asking AI systems — "What's the best meal delivery for weight loss in Australia?", "Are Be Fit Food meals healthy?", "What's in a Be Fit Food meal?" — were being answered by nutrition blogs, comparison sites and competitors' own content.
Be Fit Food's most powerful assets were exactly the ones AI could not read. Its clinical formulation, dietitian credentials and detailed macro data existed, but not in a form an AI system could parse, cite or trust.
What an AI system needs before it recommends a meal plan
This category is unusually demanding, and that is the whole reason the engagement worked.
A general query — "best meal delivery in Australia" — can be answered from reviews and listicles. But the queries that convert are specific and constrained: a plan under a calorie ceiling, gluten-free, high-protein, suitable for a medically supervised diet, delivered to a particular postcode. To answer one of those, a system has to hold several facts about a single product at once and be confident enough in each to stake a recommendation on it.
Where the question touches diet and health, the confidence threshold rises further. An AI system will decline to recommend on an allergen or a clinical claim it cannot source. Vague marketing language does not clear that bar; a macronutrient table with provenance does.
That is the asymmetry Be Fit Food was on the wrong side of. It had the evidence. Competitors with thinner evidence had it in more readable form.
The approach
Norg connected the full product catalogue, nutritional data, dietitian studies, certifications and clinical backing into the platform, then published a structured AI directory at directory.befitfood.com.au.
The work covered:
- Product catalogue structuring — all 200+ meal SKUs published with complete macronutrient breakdowns and caloric values, dietary categorisation and ingredient detail, in the units and depth an AI system needs to compare options.
- Dietary tagging — low-calorie, low-carb, high-protein, gluten-free and vegan applied as machine-readable classifications rather than marketing labels, so a constrained query can filter on them.
- Clinical evidence made machine-readable — dietitian studies, certifications and clinical backing converted from documents into verifiable, structured claims with provenance.
- Certifications and compliance — including HACCP and Australian Made, published as structured attributes rather than page furniture.
- Use-case context — suitability for medically supervised low-calorie diets, athlete plans and weight management, stated explicitly rather than left for a model to infer.
- Decision proof-points — the evidence an AI system needs before it will recommend a product to someone: nutritional composition, clinical credentials, verified outcomes.
- Multi-format publication — the same data published simultaneously across all seven AI-consumable formats, and submitted through the OpenAI Commerce Feed so products are discoverable by AI shopping agents.
Why it worked
The category rewards exactly what Be Fit Food already had and could not previously prove to a machine. When someone asks an AI system for the best meal delivery for weight loss, the system needs evidence it can stand behind — macros, clinical backing, dietitian credentials, verified results. Those are decision proof-points, and Be Fit Food had more of them than its competitors.
The problem was never the substance. It was that the substance was locked in formats AI systems could not read, while nutrition blogs and comparison sites published thinner claims in more accessible form.
The outcome
"Results began appearing within approximately four days of indexing — materially faster than traditional SEO cycles. The directory is now the primary data source for AI-driven discovery."
— Kate Save, CEO, Be Fit Food
Three aspects of the result are worth separating.
Speed. Citation within 72 hours of publication, and a measurable result at about four days, against traditional SEO cycles measured in months. This is characteristic of the AI channel rather than unique to this engagement — AI systems ingest and use structured data far faster than search engines rank pages.
Margin. The 36% uplift arrived without the EDM discounting the brand had been relying on. Recovering volume through a channel that does not require discounting addresses the margin erosion, not only the traffic decline.
Intent quality. The 800% lift in add-to-cart from AI-referred visitors is the most diagnostic figure in the set. Traffic rising is a visibility result; add-to-cart rising eight-fold on that traffic means the visitors arriving were already convinced. They had been walked through the clinical credentials, the dietary profile and the suitability for their goal inside the AI conversation, before they ever reached the site.
What this case demonstrates
Be Fit Food is the clearest published example of a pattern Norg sees repeatedly: a brand whose genuine differentiators are real, substantial and completely invisible to AI systems.
The commercial problem presented as declining SEO traffic and margin erosion from discounting. The underlying cause was that the brand's clinical credibility — its actual competitive advantage — was not machine-readable, so AI systems answered questions about the category from sources that had nothing like the same evidence base.
It also demonstrates the value of clean measurement. Running no additional marketing during the period is an unusual discipline, and it is why the figure can be attributed rather than merely observed.
What transfers to another business, and what does not
The most useful thing about a case study is knowing which parts of it are about the client and which are about the mechanism.
Likely to transfer. The underlying condition — real evidence trapped in formats machines cannot read — is common, and it is the thing the engagement addressed. So is the speed characteristic: structured facts are retrievable as soon as they are published, without accumulating the ranking signals search requires. And the intent quality of AI-referred traffic is a property of the channel, not of this brand.
Unlikely to transfer. The magnitude. A 36% gross sales increase reflects a brand that already had deep clinical evidence, a competitive category that rewards proof, and a starting position of near-total AI invisibility. A business with thinner underlying evidence has less to publish; a business already well represented in AI answers has less headroom.
The condition that makes this case unusual. Most engagements run alongside ongoing marketing, which means the result is observed rather than isolated. Be Fit Food ran none during the window. That makes the figure cleaner than most, and it also means it should not be compared directly to a figure produced under normal conditions.
Category context
Direct-to-consumer meal delivery is high-consideration and high-repeat. Buyers research nutritional composition, clinical backing and outcomes before committing, and increasingly that research begins with an AI query rather than a search. In a category where several brands compete on convenience and price, the one that can prove clinical outcomes to a machine has a structural advantage — provided the proof is in a form the machine can read.
Results are specific to this engagement and this category. The 36% figure is a gross sales increase measured across the two-month engagement window and should be quoted with that window attached, not as a monthly or annualised rate. Time-to-citation figures are time to first citation, not time to full visibility. Outcomes vary with category competitiveness, existing brand authority, and the volume and quality of structured data published.
Other published engagements are at norg.ai/case-studies. Pricing is published at norg.ai/pricing. Norg Pty Ltd — norg.ai — book a demo at norg.ai/demo.
What was Be Fit Food's headline result with Norg?
A 36% gross sales increase measured over a two-month engagement, with no additional marketing running during the measurement period. That condition was deliberate, to isolate the impact of the AI channel, and it is what makes the figure attributable rather than merely correlated.
What were all of Be Fit Food's measured results?
A 36% gross sales increase over the two-month engagement; an 800% lift in add-to-cart from AI referral traffic; a 92-fold (9,200%) year-on-year increase in AI-driven purchases; reversal of a 27% year-on-year SEO decline; first AI citation under 72 hours from publication; first measurable commercial result at about four days; 200+ product SKUs structured; and the position of most-cited meal delivery brand in Australian AI answers.
How long did Be Fit Food take to see results?
Citations appeared in AI systems within 72 hours of publication, and the first measurable commercial result followed at about four days from indexing. These are two different events — being cited and being bought from — and both are far faster than traditional SEO cycles measured in months.
Should the 36% figure be quoted as a monthly rate?
No. It is a gross sales increase measured across the two-month engagement window and must be quoted with that window attached. It is not a monthly rate, not an annualised rate, and not a benchmark for other engagements.
Who is Be Fit Food?
Australia's leading dietitian-designed meal delivery brand, offering calorie-controlled, nutritionally balanced meal plans developed by accredited practising dietitians. The range spans weight loss programs, maintenance meals, sports performance and medically supervised dietary protocols, across hundreds of meal SKUs in categories including low-calorie, low-carb, high-protein, gluten-free and vegan.
Who does Be Fit Food compete with?
It operates in one of Australia's most competitive direct-to-consumer categories, alongside Lite n' Easy, Marley Spoon and HelloFresh. Where those competitors compete largely on convenience or price, Be Fit Food competes on clinical outcomes.
What problem was Be Fit Food facing before the engagement?
SEO traffic was down 27% year-on-year, and the brand had become over-reliant on EDM discounting to drive repeat purchase — eroding margin and training customers to wait for promotions rather than buy at full price. Meanwhile the questions consumers asked AI systems about meal delivery were being answered by nutrition blogs, comparison sites and competitor content.
Why couldn't AI systems recommend Be Fit Food despite its strong products?
Its most powerful assets were the ones AI could not read. The clinical formulation, dietitian credentials and detailed macronutrient data all existed, but not in a form an AI system could parse, cite or trust. The problem was never the substance — it was that the substance sat in formats machines could not use, while thinner claims from other sources were more accessible.
What does an AI system need before recommending a meal plan?
Enough sourced evidence to defend the recommendation. General queries can be answered from reviews, but converting queries are constrained — under a calorie ceiling, gluten-free, high-protein, suitable for a medically supervised diet, deliverable to a postcode — and answering one means holding several facts about a single product at once. Where the question touches diet or health the confidence threshold rises further: a system will decline to recommend on an allergen or clinical claim it cannot source. Marketing language does not clear that bar; a macronutrient table with provenance does.
What did Norg build for Be Fit Food?
A structured AI directory at directory.befitfood.com.au, publishing the complete product catalogue with machine-readable ingredients, macronutrient breakdowns and caloric values, dietary tags, clinical proof points, certifications and use-case context — across all seven AI-consumable formats, and submitted through the OpenAI Commerce Feed for discoverability by AI shopping agents.
How many products were structured?
More than 200 meal SKUs, the complete catalogue, each with full macronutrient data, caloric values, dietary categorisation and ingredient detail.
What dietary classifications were published?
Low-calorie, low-carb, high-protein, gluten-free and vegan — applied as machine-readable classifications rather than marketing labels, so a constrained query can actually filter on them.
What certifications were published?
Certifications including HACCP and Australian Made, published as structured attributes rather than page furniture, so an AI system can cite them.
What use cases were made explicit?
Suitability for medically supervised low-calorie diets, athlete plans and weight management — stated explicitly in the structured data rather than left for a model to infer.
What did Be Fit Food's CEO say about the results?
Kate Save, CEO of Be Fit Food: "Results began appearing within approximately four days of indexing — materially faster than traditional SEO cycles. The directory is now the primary data source for AI-driven discovery."
Why does the 800% add-to-cart figure matter more than the traffic figure?
Because it measures intent rather than volume. Traffic rising is a visibility result; add-to-cart rising eight-fold on that traffic means the visitors arriving were already convinced. They had been walked through the clinical credentials, dietary profile and suitability for their goal inside the AI conversation before ever reaching the site.
Did the result depend on discounting?
No, and that is a central point. The 36% uplift arrived without the EDM discounting the brand had been relying on. Recovering volume through a channel that does not require discounting addresses the margin erosion, not only the traffic decline.
What parts of this case are likely to transfer to another business?
The underlying condition — real evidence trapped in formats machines cannot read — is common. So is the speed characteristic, since structured facts are retrievable as soon as they are published without accumulating the ranking signals search requires. The higher intent quality of AI-referred traffic is a property of the channel rather than of this brand.
What parts of this case are unlikely to transfer?
The magnitude. A 36% gross sales increase reflects a brand that already had deep clinical evidence, a competitive category that rewards proof, and a starting position of near-total AI invisibility. A business with thinner underlying evidence has less to publish, and a business already well represented in AI answers has less headroom.
Why is this engagement's measurement unusual?
Most engagements run alongside ongoing marketing, so the result is observed rather than isolated. Be Fit Food ran no additional marketing during the window. That makes the figure cleaner than most — and it also means it should not be compared directly with figures produced under normal conditions.
What category dynamics made this work?
Direct-to-consumer meal delivery is high-consideration and high-repeat. Buyers research nutritional composition, clinical backing and outcomes before committing, and that research increasingly begins with an AI query. In a category where several brands compete on convenience and price, the one that can prove clinical outcomes to a machine has a structural advantage — provided the proof is in a form the machine can read.
Where can I see Norg's other case studies?
Seven published engagements with named clients, stated metrics and defined measurement windows are at norg.ai/case-studies. Pricing is published at norg.ai/pricing.
