I was on a portfolio review last month with a brand doing about $600K a year. Apparel. The founder pulled up their promo calendar. Eleven sitewide sales in the last twelve months. Black Friday, Cyber Monday, a Click Frenzy, an Easter sale, a winter clearance, a "thank you" sale, a "we’re back" sale after a quiet July, and a few others I’ve already forgotten.

Their question to me was: "How do we discount smarter?"

My question back: "Why are you discounting at all?"

That’s where the conversation got uncomfortable. Because the honest answer wasn’t "to acquire new customers." It wasn’t "to clear end-of-season stock." It was "because we have too much stuff sitting in the warehouse and we need the cash."

That’s not a marketing problem. That’s a buying problem wearing a marketing costume.

The data nobody puts on the deck

James Hurman and Klaviyo just released a report on discounting, Sugar High, and the numbers are flat out brutal. They tracked thousands of brands globally and a deeper sample of 176 Australian brands across twelve months. The headlines:

70% of all discount events generated negative contribution profit. Among brands that discount heavily, 90% of their promotional events lost money on a contribution basis. Not failed to grow. Lost money.

Brands that ran 11 or more discount events in a year saw profit margins fall 11% year-on-year. Brands that ran five or fewer saw margins grow 8%. That’s a 19-point margin gap between brands selling broadly the same products to broadly the same customers.

And the part that should sting the most for any operator who’s run a sitewide sale: 88% of order volume during sale periods comes through discounted orders, and 68% of those orders are placed by returning customers. At full price, only 11% of orders are from returning customers.

Translation: when you put your site on sale, you’re handing a discount to the customers who would have paid full price anyway. And you’re selling them your hero products, not your dead stock.

Class A and B products account for 90.26% of revenue during sitewide sales. Class D, the slow-moving stuff you actually wanted to clear, accounts for 0.2%.

Source: Klaviyo × James Hurman, Sugar High. ProfitPeak data across 176 brands.

Two-tenths of one percent. So much for the clearance theory.

Why this happens

The whole report is worth reading, and you can read it here. But the bit that hit hardest for me was the case study with Justin Hillberg at Culture Kings. Their situation is different scale, public-market pressure, much bigger inventory exposure, but the mechanic is identical to what I see across DTC brands at $500K, $1M, $5M. Hillberg’s words: "The promotional cadence was a symptom of a procurement problem and lack of inventory management discipline."

That’s the line. Read it again.

Most brands I see running monthly promotional events aren’t doing it because they sat down and decided promotional cadence was their growth lever. They’re doing it because they over-bought twelve weeks ago, or six months ago, or last season. The sale isn’t the strategy. The sale is the symptom.

The sale isn’t the strategy. The sale is the symptom.

And once you start running sales to clear cash flow tension, you train two things at once. You train your customers to wait. And you train your team to use the discount lever instead of fixing whatever’s broken upstream. Both habits compound. The customers wait longer. The team reaches for the lever faster. And every cycle drops the reference price your customers carry in their heads a little lower.

How to tell which problem you actually have

This is the diagnostic question I’d run before changing anything in your promotional calendar.

Pull every discount event from the last twelve months. For each one, answer honestly: what was the real reason we ran this? Not the marketing-deck reason. The actual reason.

Three buckets:

Strategic discounts. Acquiring a specific customer segment. Reactivating a lapsed cohort. Marking a moment that genuinely matters to the brand. Moving genuine end-of-life stock. These are real jobs and a well-designed discount can pay for itself.

Calendar discounts. "It’s Click Frenzy and our competitors are running one." "It’s Black November so we have to." "Mother’s Day weekend, every year." These are reactive. They might be profitable, but mostly they’re treadmill events. Everyone runs them. Nobody really benefits.

Cash flow discounts. "We need to move stock to fund the next order." "AP is tight." "We over-bought and we need to recover some of it." This is the symptom category. The discount is doing the job your inventory planning should be doing.

If more than a third of your last twelve months sit in the third bucket, your discount problem is downstream. Fixing the promotional calendar won’t fix it. You have to fix the buying.

The crawl, walk, run for getting out

Going cold turkey doesn’t work. Hurman’s data shows brands that yank discounts year-on-year see all-buyer GMV growth roughly halve. Customers trained to wait don’t suddenly start paying full price. They just wait longer.

So this is sequenced.

Crawl: get honest about your buying. Before you change anything customer-facing, look at your last three POs. Did you buy to a forecast, or did you buy to a target growth number you hoped to hit? There’s a difference. The first is grounded in your actual sell-through. The second is wishful thinking that ends in a warehouse problem six months later. If your buying has been hope-based, that’s the first thing to fix. Tighten your demand planning. Buy closer to actual sell-through. Use Shopify’s reporting plus your past twelve months as the baseline, not the optimistic forecast in your investor deck.

Walk: separate strategic from reactive promotions. Look at your next twelve months of planned promos. For each one, write down the job. If you can’t define a job beyond "it’s that time of year," that’s a cut candidate. You don’t have to cut it this quarter. But map which events are pulling weight and which are just calendar reflexes. Then start reducing the reflex ones one at a time, replacing them with smaller, more targeted plays. A targeted lapsed-customer reactivation does more for your margin than a sitewide 20% off, and it doesn’t train your full-price buyers to wait.

Run: rebuild the metrics around margin, not revenue. This is the Hillberg play, and it’s the hardest one. If your team is still tracking weekly revenue and unit volume as the primary numbers, they will quietly find ways to keep the discount engine running because those are the numbers that respond fastest. Add contribution profit per order. Add gross margin dollars. Add the split between full-price actives and discount actives in your customer cohorts. Get the team celebrating full-price growth even when total revenue is flat or down for a quarter or two. The shrink is part of the work.

Where Klaviyo actually helps here

One thing worth pulling out of the Hurman report. The 68% returning-customer stat is the one I’d put on a sticky note. If two-thirds of your discounted orders are going to people who would have paid full price, you don’t have a discounting problem so much as a segmentation problem. You’re sending the same discount to everyone because your data isn’t telling you who’s actually at risk of churning versus who’s just waiting for a deal you’ve trained them to expect.

Klaviyo is the obvious lever here. Lapsing-customer flows. Cohorted promotions to customers genuinely at risk, not your top buyers. Replenishment reminders that don’t need a discount attached. Most brands I work with have Klaviyo and are using maybe 30% of what it can do. The other 70% is exactly the infrastructure that lets you discount less, not more.

The honest punchline

Discounting feels like marketing because it shows up in marketing channels. But the brands stuck in the loop almost always have a buying problem, a forecasting problem, or a cash flow problem that the discount is papering over.

You can keep discounting your way out. The Hurman numbers tell you where that lands. Or you can ask the harder question first: what’s actually driving this?

Fix that, and most of the promotional calendar fixes itself.