What we learned building a hospo-tech platform in Australia in 2025–26

There is a statistic that stopped me cold when I first came across it. According to the Australian Financial Review, the average Australian restaurant makes approximately 60 cents in profit on every $30 spent by a customer. Two percent. Not after a bad month — as a structural reality, baked into wages, rent, COGS, and insurance. An industry that employs 10% of Australia's workforce and sits at the centre of how we socialise, celebrate, and feed ourselves, running on margins thinner than a crêpe.

When you understand that number, the rest of the last eighteen months make sense. Why operators are exhausted. Why are they making decisions they know aren't good for their businesses? Why a platform that offers to fill their slow periods in exchange for 30–40% of the bill — effectively zeroing out that 60 cents and then some — can still find 5,000 venues willing to sign up. Desperation is a powerful sales tool.

We started Rydra because we saw an industry being systematically extracted from, and we thought there was a better way to build technology for it. What follows is what we actually learned — about the market, about the competitors, about the regulatory environment we're building into, and about what it means to try to build something genuinely aligned with the venues it serves.

What the operators told us

In the early days, before a single line of code and before a single venue signed up, we spent a lot of time listening. Not pitching — listening. We sat across from venue operators, café owners, bar managers, and group operators, and we asked them what was actually going on.

The pattern that emerged was consistent enough to be uncomfortable. Most operators were not failing for lack of customers or for want of a good product. They were failing because every growth lever available to them extracted value rather than creating it. Uber Eats at 30% commission. DoorDash at up to 35%. EatClub is asking them to hand 30% off the bill to every customer who walks in during slow periods, plus a 10% platform commission on top. The industry response to the Broadsheet/Square survey of 69 operators in mid-2025 put numbers to what we were hearing: 88% faced lower profit margins year-on-year. 71% had seen revenue decline. Three in four owners were spending more time on operations than the year before.

One operator we spoke to — a well-run café in Melbourne, consistent trade, good reviews, not doing anything obviously wrong — described their situation with a clarity I haven't been able to forget: "We're busy. We just make no money." Another described the platform fees as "paying to exist." A third said, with no self-pity, "I can tell you exactly how many Uber Eats orders it takes to cover my rent. The answer means I need more orders than I can physically do in a day, and I'm still not profitable on any individual one."

These were not failing operators. These were the people the industry held up as examples of what to do, running at the edge of viability because the technology supposedly helping them was, in most cases, helping itself first.

That's where Rydra starts.

The flat-fee decision

The first major decision we made — and the one that shaped everything else — was to charge a flat $1.99 platform service fee directly to the consumer, and nothing to the venue. No commission. No percentage of the order. The venue keeps 100% of every dollar spent on their food.

This was not a difficult decision intellectually. The maths are not complex. A venue doing 50 orders a day at an average of $40 per order, processed through a 30% commission platform, loses $219,000 a year in platform fees. At $1.99 flat, the consumer-side cost for that same volume is $36,330 — a difference of $182,670 that stays in the venue's P&L. At 2–5% margins, that's the difference between a business that survives and one that doesn't.

What made it a harder decision was the business model implication. You can't take a percentage of every order. Your revenue doesn't scale automatically with venue success. You have to be good enough at what you do — valuable enough to enough venues and consumers — that the flat fee volume makes the model work. That's a more honest arrangement, and it's a harder one to build.

We made peace with that early on. If we can't earn our place at $1.99 without extracting commissions, we don't deserve to exist.

The EatClub problem — and why it matters

I want to talk about EatClub directly, because I think the industry deserves a frank conversation about what the discount model actually does to venues.

EatClub is a Melbourne-founded startup that has raised over $45 million — most recently a $27 million Series B in February 2026 — and claims 5,000+ venue partners across eight Australian cities. They describe their product as "dynamic pricing." Their marketing is sophisticated. Marco Pierre White is a shareholder and brand ambassador. The pitch to investors is compelling.

The pitch to operators is this: use our platform to fill your slow periods by offering time-limited discounts of up to 50% to our user base of 2 million+ deal-seeking consumers. When a customer pays through EatClub Pay, the venue nominally receives the full bill amount. EatClub covers the discount gap. In exchange, venues pay subscription fees and, per a verified Trustpilot review from a venue owner in November 2025, around 10% commission on top of the discount they're offering.

That reviewer — Kevin Hope, a named, verified operator — wrote something that I think every venue considering EatClub should read in full. I'll paraphrase the substance: venues are pitched on the idea that 30% discounts plus 10% commission still leaves a healthy margin. Over time, the venues that join become dependent on the platform's customer traffic. The platform reinvests its profits into acquiring more consumers — consumers who are loyal to the deal, not the restaurant. Margins are destroyed quietly, deal by deal, until the only visible venues on the platform are those who offer the deepest discounts, and the only customers coming through the door are the ones who won't return at full price.

He ended with: "I know a well-known local café that is extremely busy every day, yet makes no profit at all. The business is now struggling and has been put up for sale."

EatClub currently holds a 2.1 out of 5 rating on Trustpilot, with 76% of reviews being one star. The company has claimed the profile. It hasn't replied to the negative reviews.

The dynamic pricing language is worth examining closely. Yield management worked for airlines because a seat on a Tuesday afternoon flight has zero marginal cost for the airline — the plane is flying regardless. A restaurant kitchen is different. Every meal served at 50% off has the same COGS, the same labour cost, the same energy cost. The comparison flatters the model and obscures the maths.

The deeper issue is what discounting does to customer quality. Academic research on this is unambiguous. A Rice University study found that only 19.9% of discount-acquired restaurant customers return at full price. A separate study found that customers who visit for the first time at full price are more than twice as likely to make a second visit compared to those acquired through a deal. The University of Maryland analysed 2,000 restaurants and found that daily deal platforms reduce perceived food quality AND service quality in consumer reviews — even for restaurants that run deals well. You can measure the reputational damage.

EatClub's own power users describe finding the same restaurants cycling through its platform as through Uber Eats, DoorDash, DineOut, First Table, and the Entertainment Book. That's not venue acquisition intelligence — that's the same pool of deal-seeking consumers being passed between platforms, none of whom are being converted into loyal customers of the venues they visit.

The cautionary precedent here is Groupon. In 2011, Groupon was valued at $17.8 billion — the fastest company in history to reach a $1 billion valuation. It rejected a $6 billion acquisition offer from Google. By 2023, its market cap had fallen 99.4% to $103 million. Revenue dropped 80%. It issued a going-concern warning. The mechanism of its failure is well-documented: it trained restaurant customers to wait for discounts, cannibalised existing full-price customer bases, offered venues no control over volume, and gave them no customer data to build loyalty. The deal-seeking audience it built had no relationship with the venues it ate at.

EatClub has made genuine improvements on the Groupon model. Venues control their own timing and discount depth. The EatClub Pay mechanism removes the awkwardness of presenting a voucher. The Obee booking integration is genuinely useful. And unlike Groupon, they do appear to be investing in platform development rather than simply extracting and exiting.

But the fundamental economics haven't changed. When you train customers to expect 30–50% off, you're not building loyalty to a venue — you're building loyalty to the discount. When you tell venues that quiet periods are the right time to fill with deal-seekers, you're obscuring the fact that deal-seekers often use the entire bill as their tipping base, order one course instead of two, and represent a net negative per-cover compared to full-price customers who chose the venue on its merits.

We built Rydra to do something different. Not to fill slow periods with discount-motivated customers who will never return. But to help venues understand who their customers are, reach them directly, bring them back at full price, and build the kind of retention data that makes a business actually defensible.

How the RBA surcharge ban shaped our product

We launched Rydra in an environment where the regulatory ground was already shifting, and one change in particular shaped our fee architecture in ways I'm glad about: the RBA's decision to ban card payment surcharges from October 1, 2026.

The ban covers all eftpos, Visa, and Mastercard transactions. It is estimated to save consumers $1.6 billion annually. For hospitality specifically — an industry where surcharges of 1–1.5% have become standard — it eliminates what was functionally a pricing mechanism that operators depended on to partially offset their card processing costs.

The Australian Restaurant & Café Association argued, with some justification, that venues will now absorb processing costs directly rather than passing them on. CEO Wes Lambert called the announcement "a sad day for hospitality." Eighty-one percent of hospitality businesses currently surcharging face a direct P&L impact.

For Rydra, this was a product design clarification. Our $1.99 platform service fee is charged to the consumer and is payment-method agnostic — it isn't a surcharge on card transactions. It's a transparent, upfront fee for platform access, consistent regardless of how the customer pays. We're structurally compliant with the new environment before it takes effect. And because the ban is arriving six months after this article publishes, platforms that relied on surcharge pass-through are now in a scramble to restructure.

The broader lesson here is one we learned early: building into the regulatory environment rather than around it. The surcharge ban was flagged years before it landed. Platforms that built their economics around it were building on borrowed time. Gig worker reforms ($31.30 minimum hourly rate for delivery workers from July 2026) are doing the same to delivery-dependent platform models. Every time a regulatory change makes a delivery aggregator more expensive to use, the case for owned direct ordering becomes stronger.

Building the predictive analytics system

The third pillar of Rydra — after the flat fee and the owned ordering channel — is what we call ClubRydra: a behavioural analytics system built on a Markov Chain model of customer state transitions.

The insight behind it is this: hospitality analytics is almost entirely backwards-looking. Most platforms tell you what happened — how many covers, what the average spend was, and how many repeat orders came through. What they don't tell you is what is about to happen. Which customers are drifting toward churn? Which ones are on the verge of becoming regulars? Which ones, if reached in the next seven days, are likely to return?

The system we built treats every customer interaction as a movement between states — browsing, viewing a deal, adding to cart, ordering, returning, or exiting — and calculates the probability of the next state based on observed behaviour across thousands of users. The output is a Rydra Loyalty Score for each customer at each point in their journey, paired with a churn probability. A customer who ordered once and hasn't returned in 18 days has a quantifiable probability of never coming back. The question is whether you know that and act on it, or whether you find out retrospectively when you count last quarter's covers.

We built this not because behavioural modelling is intellectually interesting (though it is) but because the economics of retention are so much more powerful than the economics of acquisition. Bloom Intelligence's 2025 data puts the lifetime value of a one-time visitor at $26. A customer who comes back just once jumps to $345. A regular reaches $685. The gap between one visit and two is a 13x increase in lifetime value. The entire justification for a loyalty and predictive analytics system lives inside that multiplier.

The system improves as more data flows through it. Every order placed, every cart abandoned, every return visit or dropout sharpens the transition probabilities. This means growth doesn't just increase revenue — it increases the intelligence of the platform. The competitive advantage compounds.

What the market taught us about building in the cold

The hospo-tech funding environment in 2025–26 is dramatically different from 2021–22. The era of growth-at-all-costs is over. Every rapid grocery delivery startup in Australia has been wiped out — Milkrun collapsed despite $86 million invested; Deliveroo exited in 2022, losing $33 million; Menulog shut its entire Australian operation in November 2025 after 19 years and roughly 24% market share. The duopoly of Uber Eats (~54%) and DoorDash (~15%) that remains is under pressure from gig worker cost reforms that will compress their economics from July 2026.

The me&u/Mr Yum merger is instructive from the other side. Two well-funded QR ordering platforms — competing for the same venues, running losses on both sides — merged in late 2023 and reached profitability in the second half of 2025. Kim Teo, the CEO, described the path with unusual clarity: "Not because we slashed everything, but because we balanced cost control with product progress." That's the template for surviving this environment. Build something useful. Charge what the economics require. Don't rely on the theory that growth will eventually fix unit economics that are broken by design.

We have tried to internalise that lesson at every stage. The flat fee is low enough to be instantly acceptable to venue operators and consumers. It's high enough, at volume, to sustain a business. The math has to work at the unit level before scale is relevant.

The thing about alignment

The last thing I'll say is about something you can't quantify, but that shapes every decision in a business like this: whose side are you on?

Every platform that charges commission is structurally in conflict with the venues it serves. When the venue does more volume, the platform takes more. When the venue raises prices, the platform takes more. The platform is incentivised to maximise order volume regardless of margin, because margin is the venue's problem and volume is the platform's income. That's not a conspiracy — it's just what the incentives produce. It's also why the discourse around these platforms is so consistently uncomfortable: the things platforms say in their marketing ("we help venues grow," "we're a tool for profitability") are not quite false, but they paper over the structural misalignment that makes them, at best, an expensive growth channel and, at worst, a slow margin erosion.

A flat fee changes the incentive structure. Our fee doesn't grow when your prices grow. We don't earn more when you take more orders. We earn the same $1.99 whether your order is $15 or $85. That means our growth model is about adding venues and consumers — not about extracting more from the ones we already have. That's not a marketing position. That's what the incentives actually produce.

The operators we've spoken to respond to this difference immediately. They've been pitched by enough commission-based platforms to understand the model. The conversation changes when the fee structure changes.

Where things stand

This is a partial account. Building anything real is messier than the version you can tell in a blog post, and more dependent on people — on the operators who gave us their time before we had anything to show, on the team members who took a risk, on the early consumers who used the platform when it had twenty venues and a lot of rough edges.

What I can say with confidence is this: the problem we set out to solve is real, the extractive models we're competing with are visible and documentable, and the regulatory environment is moving in directions that reward platforms with clean economics and penalise those that relied on surcharges, hidden commissions, and margin arbitrage.

Australian hospitality employs nearly a million people. It generates $66.3 billion in annual turnover. It is one in ten businesses in this country. And it is, right now, running at 60 cents of profit per $30 in revenue, being squeezed by platforms that take a percentage of every transaction, offering discounting tools that train customers to never pay full price, and facing a regulatory shift in six months that will eliminate a fee offset they depended on.

There is a better way to build technology for this industry. We are trying to build it.

Rydra is a hospo-tech platform connecting consumers to hospitality venues through a flat $1.99 platform service fee — charged to the consumer, not the venue. No commissions. No margin extraction. Owned customer data and predictive analytics built in. If you're an operator who wants to talk, or a journalist covering the space, we're at getrydra.com.

Sources: CreditorWatch Business Risk Index 2025 · ASIC Insolvency Statistics FY24-25 · Broadsheet/Square Operator Survey, May 2025 · Australian Financial Review restaurant margin analysis · RBA Conclusions Paper, March 2026 · Fair Work Commission gig worker minimum standards submission, November 2025 · EatClub Trustpilot (au.trustpilot.com/review/eatclub.com, accessed April 2026) · SmartCompany EatClub Series B coverage, February 2026 · Bloom Intelligence State of Restaurant Guest Retention 2025 · Rice University study, Dr Utpal Dholakia, 2010 · University of Maryland INFORMS study, Mejia, Gopal, Trusov · Wharton School "The Death of the Daily Deal" · TechCrunch Groupon IPO/decline coverage · Information Age "Menulog to quit Australia," November 2025 · Rydra ClubRydra Behavioural Intelligence Framework, internal research 2026