Almost every brand offers online retail discounts. Most discount more than they did 5 years ago and more than they did last year. And most plan to discount even more in the future.
Black Friday has expanded into Black November. The 20%-off welcome offer has become the price of admission to a customer's inbox. The "limited time" sale has, for many brands, become the only time.
I've spent the last several months digging into what that pattern is actually doing to the brands caught in it.
The findings don’t make for comfortable reading. This blog is based on Sugar High!, an independent research paper produced in partnership with Klaviyo that draws on Klaviyo's global brand dataset and ProfitPeak's analysis of 176 Australian ecommerce brands.
The brands discounting hardest are growing the slowest
Conventional wisdom says heavy discounters are buying growth with margin, trading profit for top-line velocity. But the data doesn't support that. In fact, it shows that they're actually trading profit for less top line.
Brands with low discount rates grew GMV 12% year-on-year in Klaviyo's analysis of thousands of brands from Q1 2025 to Q1 2026. Deep discounters grew 6%. That's a gap of almost 100% in growth rate between brands selling broadly the same products to broadly the same customers. The only meaningful difference is how often they put things on sale.
The profit picture is uglier still. Low discounters posted 8% margin growth. Deep discounters—brands running 11 or more promotional events a year—saw their profit margins fall 11%. That's a 19-point margin gap. Every quarter, the dashboard says the sale worked. Every year, the books say it didn't.
It's worth pausing to understand why this happens. A heavy discounter is pulling forward future demand at a lower margin and training their highest-intent customers to wait. The growth that shows up in any single sale period is real, but the growth that shows up over 12 months is what's left after the cannibalisation, the pull-forward, and the customer retraining all net out. The annual chart is the honest one.
You're not acquiring new customers. You're discounting your existing ones.
Ask any marketer why their brand runs a sale and somewhere in the answer you'll hear the word "acquisition." Discounts pull in new customers. People who weren't going to buy at full price decide to buy at the lower one. The funnel widens.
That's the theory, but the data says something else entirely.
In the ProfitPeak dataset, 88% of order volume in the 12-month period flowed through discounted orders. Only 12% came through at full price. So whatever discounting is doing, it's doing it to almost everything you sell. And inside that 88%, the customer mix should give marketers pause: 68% of discounted orders were placed by returning customers. The equivalent figure for non-discounted orders is 11%.
That's a 57-point swing in repeat-customer share, depending only on whether the price was full or reduced. In other words, your sale isn't widening the funnel. It's subsidising the customers who were already in it.
The revenue you capture during a sale is borrowed from the weeks after it
A sale period feels like growth in real time. The dashboard lights up. Conversion rates jump. Cart abandonment falls. The team congratulates themselves.
Now look at the 4 weeks after the sale ends.
In the ProfitPeak data, topline revenue in the 2–4 weeks following a discount period fell by an average of 27% against the brand's normal baseline, not against the inflated sale week. Your brand sells less in the weeks after a sale than you would have sold if the sale had never run.
This is the most well-documented effect in promotional pricing, and it's not a mystery. Customers who were going to buy anyway brought their purchase forward to capture the discount. Once the sale closes, the pipeline that would have fed the next 4 weeks is empty. The revenue you booked during the sale was borrowed from the revenue that was coming next, and you paid a margin penalty for the privilege of borrowing it.
Plotted on a chart, the brand's revenue stops looking like growth and starts looking like a heartbeat: spike, crash, slow climb, spike, crash.
Your sitewide sale is discounting your hero products, not clearing your slow stock
When a brand runs a sitewide sale, the internal justification almost always has two parts: it'll move slow-moving stock, and it'll introduce customers to lines they wouldn't otherwise have tried.
So which of those two things actually happens?
Neither.
In the ProfitPeak data, Class A and B products—bestsellers and core range—accounted for 90% of all revenue during sitewide discount periods. The long tail of products brands most want to clear accounts for 0.2% of revenue, or two-tenths of one percent. And that’s during the very promotional periods designed to do the clearance work.
What's actually happening here is straightforward: customers don't browse a sale, they search a sale. They come in with the products they already know, already trust, and already want. They check the price, they click “Buy,” and the clearance racks stay full. Your hero products go out the door at 20–40% less than they were worth that morning.
And then there's AI
In 2026, heavy discounting may feel more necessary than ever. The cost of living crisis has been real. Deloitte's 2026 Consumer Products Outlook reports that 47% of consumers globally now qualify as "value seekers," meaning people who routinely make convenience sacrifices, cost-conscious choices, and deal-driven purchases.
When a marketer pushes back on this research and says "We have no choice, the consumer is asking for it," they're not wrong about the consumer. But they are wrong about the right commercial response.
The second force shaping this environment is newer and hasn't been fully thought through yet. AI-assisted shopping has moved from curiosity to mass behaviour in about 18 months. McKinsey reports that 44% of AI-powered search users now say AI is their primary source of insight when shopping. AI agents drove 70% more completed shopping tasks during Cyber Week 2025 than the year before.
A reasonable conclusion would be: the future is agentic, and the AI agents will all hunt for the best price. That makes discounting a strategic necessity in the AI era.
I'd argue the opposite.
AI models are not purely rational price-comparison engines. They're trained on datasets that reward signals of authority, consistency, third-party validation, and reputation. Brand signals matter as much as product credentials in how AI selects what to recommend.
A brand whose only digital fingerprint is "30% off this weekend only" is teaching the model that they’re a price brand, not a quality brand. The AI will recommend that brand on price comparison queries, not on "best in category" queries. And a smaller, cheaper, faster competitor will increasingly serve the price comparison query.
The cost-of-living crisis is making the short-term case for discounting feel inescapable at the same time the rise of AI shopping is making the long-term case against it more dangerous.
Going cold turkey doesn't work, either
If you're reading this thinking the answer is to stop discounting immediately, the data has a warning for you, too.
In Klaviyo's analysis of thousands of brands over the past year, brands that decreased their discount rates saw all-buyer GMV growth roughly halved, in both ANZ and global markets. New buyer growth turned negative.
That’s because customers who have been trained to wait for a sale don't suddenly start paying full price when the sale doesn't arrive. Instead, they wait longer while your brand's growth slows and your CFO panics. Your brand relapses, runs another sale, sees growth recover in the short term, and trains the wrong behaviour even more deeply into your customer base.
This is the addiction loop. And it's why most brands, once they're in it, never get out.
The advice is to taper, not quit cold turkey. Slowly reduce both the depth and frequency of discounting. Replace sitewide events you’re running because “it’s that time of year” with targeted, purposeful ones—discounts with a defined job, a defined budget, and a defined measure of success.
The Culture Kings reality check
Justin Hillberg took over as ANZ president of Culture Kings in January 2025 and inherited a premium streetwear brand that had spent several years doing exactly what this blog warns against: running monthly liquidation events, always-on bundle deals, and a sitewide promotion at every market moment the calendar offered.
Under the previous model, discounting had become so embedded in the business that the promotional cadence was actually a symptom of a procurement problem: the business had bought too much inventory and used promotions to move it, month after month.
Hillberg's first move was to work out why the business was discounting at all. The answer was upstream of marketing: Culture Kings had over-bought inventory chasing growth targets that didn't arrive, and promotions had become the release valve. Once that was diagnosed, the fixes followed.
Two years into the rebuild, sitewide events are effectively gone. Private-label product has grown from a minority of revenue to over 50%. Gross margin has grown MoM through the first half of 2026.
The comp-against-last-year numbers are messier, and Hillberg is straight about that: "It's not apples for apples when we look at comp. I feel like a bit of a broken record internally, just focusing on the quality metrics. We sold a lot of units this time last year, and we're not selling as many units this year"
But the direction is clear. "It's a more profitable way to do business,” Hillberg explains. “Our active customer cohort is much healthier."
That's what coming off the discount habit looks like in a real P&L: steadier, more consistent, and, on every measure that matters, healthier.
3 things to do now
The brands in this data that discount with discipline share one approach: they treat each promotional event as a budgeted investment with a defined return, not a sales-pipeline rescue.
- Audit what your discounts are actually doing, not what they were designed to do. For each promotional event in the last 12 months, ask: did it acquire new customers, or did it give existing customers a discount they didn't need? Did it lift the 4 weeks around it, but cannibalise the 4 weeks after? Did it move slow stock, or discount our hero product to our best customers? If you can't answer those questions for every event, you don't have a discounting strategy. You have a discounting habit.
- Set a budget for the loss, not the lift. The brands that discounted well in our dataset treated every event the way a good retailer treats a sale: a defined event, with a defined job, a beginning, an end, and a measure that tells you afterward whether it worked.
- Invest the margin you save in your brand. Brand strength is what protects margin in both the human economy and the AI economy. Every dollar you redirect from a discount that was about to lose money on contribution profit, into brand marketing that builds future demand, is a dollar moving from the loan account to the asset account.
Your brand is your asset. A discount is a loan against it.
Stop borrowing.
James Hurman is co-founder of Tracksuit and Previously Unavailable, and one of the world's best-known experts on marketing effectiveness. This blog is based on Sugar High!, an independent research paper produced in partnership with Klaviyo that draws on Klaviyo's global brand dataset and ProfitPeak's analysis of 176 Australian ecommerce brands.
Related content
