Regulators keep finding the same slippery discount
Three things happened this year that show the direction of travel with discounting.
- India passed a rule about how discounts get calculated.
- Amazon quietly tightened how it validates the price next to the strikethrough.
- Beijing summoned two of its biggest platforms and told them to stop.
None of these is connected. There are different regulators, different countries, and different reasons for showing up. But look at what each one is actually going after, and they are all circling the same slippery little trick.
The inflated “was” price
Let’s start with India. From 1 January 2027, the Consumer Protection (E-Commerce) Amendment Rules require that any advertised price cut be measured against the lowest price a product was actually offered at in the preceding 30 days.
That sounds technical, but it's really just a rule against pretending. You cannot mark something up for three weeks, so the 40% off looks bigger. The “before” price has to be a price that was real, recently.
Amazon reached a similar conclusion. From 23rd April 2026, a seller’s reference price only counts if customers actually bought at that price as the featured offer, or if another retailer was recently charging that price.
From 18th May, if more than half of a product’s last 90 days' sales were below its normal non-promotional price, Amazon recalculates the “was” price using all of that promotional history, not just the flattering bit. Fail either check, and the discount badge simply does not show. No warning email, no grace period. The strikethrough just disappears, and Amazon has said the promotional frequency for that product can be affected for a rolling 90-day period afterwards.
Then, Beijing goes after the most obvious version of the same problem. Regulators called in Alibaba and JD.com over “618”, the mid-year sales event, for what they described as:
- false promotional claims
- non-transparent business practices
- a failure to disclose sellers’ information
- a failure to disclose the funding split between the platform and the merchant behind a headline discount
One seller’s example did the rounds: a street food set priced at 19.8 yuan that netted the merchant 2.58 yuan after the platform’s cut. Regulators also flagged platforms quietly absolving themselves of liability when those disputes went wrong, and told them plainly to compete on service and innovation, not subsidy wars that bankrupt the people actually selling on them.
Nobody banned discounting
It’s worth being precise about what actually happened here, because it is not what the headlines make it sound like.
None of these rules says you cannot discount. India still lets you cut 40% off the genuine price; Amazon still runs deals; and Beijing didn’t end in “618”; it is still happening every June.
What each of them is squeezing is the same specific move: manufacturing the “before” number so the discount looks bigger than it is. That is a narrower target than it sounds, and it is worth checking whether it applies to anything currently in your pricing.
Picture two versions of the same product page:
Version one: a jacket has been £80 for most of the last two months, it goes on sale at £64, a genuine 20% off.
Version two: the same jacket gets bumped to £120 for three weeks, then “discounted” back down to £80, a fake 33% off a price nobody ever actually paid. The same final price on both, but one of them is now a compliance issue across three markets.
That is the whole trick, and it turns out it is a fairly small one to close. Which is presumably why three unconnected regulators managed to close it within a few months.
This was already a losing bet
Here is the thing about the inflated reference price. Even where nobody was regulating it, it mostly didn't work.
Les Binet’s research with Nielsen puts around 84% of price promotions in the loss-making column. Most discounting was already destroying margin before anyone in India, Seattle or Beijing wrote a rule about it. The regulation is not creating a problem. It is making an existing one harder to hide.
Meanwhile, BCG’s research on the alternative points in the other direction: personalised offers, aimed at shoppers who actually need them, generate roughly 3x the ROI of a blanket promotion. The gap between “discount everyone the same amount off a manufactured baseline” and “discount the people who need it, based on what they were actually going to do” was already large. It is just now also the gap between compliant and non-compliant.
What still works once the trick is gone
None of this means discounting stops converting. Recent research from Salsify puts limited-time discounts as the single biggest driver of purchase decisions, cited by 62% of shoppers, ahead of bundles at 45% and personalised or loyalty-based offers at 39% each.
Look closely at that list and notice what it does not depend on. A limited-time discount works because the deadline is real, not because the “before” price was inflated to make the saving look bigger. A bundle works because two things together genuinely cost less than two things apart.
A loyalty discount works because it rewards a customer you can already see. None of the tactics people actually respond to need the manufactured comparison. It was always possible to discount well without it. Most retailers just found the fake version easier to switch on at scale.
Even something as basic as free shipping makes the point. Long before any of these regulations, UPS found that 52% of shoppers had added items to a basket purely to qualify for free delivery. Nobody needed a fictional reference price to make that work. It just needed to be a real, useful thing to offer.
Four things worth doing
1. Audit your reference prices before your regulator or your platform does it for you
If your “was” price would not survive Amazon’s own new checks, it is worth knowing that now rather than finding out mid-Prime Day.
2. Ask whether your current discount depth is a real incentive or a manufactured one
A genuine 20% off a genuine price and a fake 60% off an inflated one can look identical on the page. Only one of them survives what is coming.
3. If you sell into India, put 1 January 2027 in the diary properly
That is not guidance. It is a hard date with a specific calculation behind it.
4. Build your promotions around a tactic that survives the audit, not just a number that survives this quarter
Bundles, loyalty tiers, and offers aimed at a shopper you can see are genuinely about to leave all clear the bar these rules are setting, because none of them was relying on the inflated baseline in the first place.
Where this is pointing to
The tactics getting squeezed are the ones that relied on the comparison doing the selling: a big number next to a bigger, fictional one. The tactics that hold up are the ones that were never played in that game. A loyalty discount, a bundle, an offer aimed at someone who was genuinely about to leave, none of those need an inflated baseline to look good.
They just need to be real.
That has always been the argument for predicting who actually needs an incentive, rather than broadcasting one price cut to everyone and hoping the maths works out. It is a slower way to build a promotions strategy than typing a larger number in strikethrough, and it was already the better bet on margin alone.
It turns out regulators are starting to agree, market by market, without comparing notes. Not because they read the same research as Les Binet or BCG. Because the fake version was always the one with something to hide, and hiding things in three different jurisdictions at once was never going to stay easy.
