Does a "Bestseller" badge make a product sell, or does it simply point at a product that was already selling? Most stores never ask, because the dashboard gives an answer that feels like proof: badged products convert at 4.1 percent, unbadged ones at 2.6, so the badge "works".
That comparison measures the product first and the badge a distant second. Badges are a small piece of interface, and they deserve a place on most listing pages.
But the way they are usually evaluated makes them look far stronger than they are, and the decisions that follow (more badges, louder badges, badges on everything) tend to cost more than they earn. The argument below runs through what badges signal, why the standard measurement misleads and what a badge policy looks like when it has survived a fair test.
The 60 percent lift that was mostly the product
The pattern is easy to recognise. A store picks its 20 top-selling products, adds a "Bestseller" label, and a month later compares their conversion with the rest of the catalogue. Badged products convert 60 percent better. The number goes into a quarterly review, and someone proposes adding "Trending" and "Staff pick" labels to the next 40 products.
The flaw is selection. Those 20 products were chosen because they already converted best: better photos, better prices, more reviews, or simply a better fit for what visitors want. Their conversion before the badge was probably already far above average. Comparing badged with unbadged items after launch mixes two effects, the product's own pull and the badge's extra push, and attributes all of it to the badge.
Even a before and after comparison on the same 20 products is weak. Bestseller lists are recalculated, so products that earn the badge are often at a sales peak, and peaks tend to fall back. Seasonality, a new campaign or a price change in the same month muddies it further.
None of this means badges do nothing. There is a sound mechanism behind them. Robert Cialdini's book "Influence" (1984) described social proof: people facing uncertainty copy what others chose. A shopper scanning 48 near-identical products on a category page has exactly that uncertainty, and a badge lowers the cost of choosing. On marketplaces the mechanism is formalised. Amazon's "Best Seller" badge goes to the top-ranked item in a category by Best Sellers Rank, which Amazon says is updated hourly, so the label carries real information about recent sales.
The badge effect is real; it is just smaller than the naive number and it depends on what the badge says.
What New and Bestseller badges actually tell a shopper
Different badges answer different shopper questions, and they move different customers. Grouping them as "badges" hides most of what matters.
| Badge | Question it answers | Shopper it helps | Main risk | Where it tends to add most |
|---|---|---|---|---|
| Bestseller | "What do other people pick?" | First-time visitors, uncertain buyers | Pulls sales from other items without growing the category | Large categories with many similar products |
| New | "What haven't I seen yet?" | Returning customers, fans of the brand | Shown too long, it stops meaning new | Fashion, beauty, collectibles, frequent drops |
| Low stock | "Should I decide now?" | Shoppers already close to buying | Misleading if stock is not really low | High-intent product pages |
| Staff pick | "What would an expert choose?" | Buyers who distrust popularity | Reads as advertising if overused | Specialist stores with visible expertise |
| Sale or price drop | "Is this a good deal?" | Price-sensitive shoppers | Trains customers to wait for discounts | Clearance and seasonal ranges |
The two badges in this guide's title work in almost opposite directions. A Bestseller badge reduces risk: it tells a nervous visitor that hundreds of people made this choice. It is most useful to new visitors and least useful to loyal customers, who already know the range. A New badge creates curiosity and is most useful to returning customers, because a first-time visitor sees everything as new anyway.
That difference changes how you should measure them. A Bestseller badge should be judged on new visitor conversion and on whether it grows the category or just shuffles demand. A New badge should be judged on returning visitor engagement: clicks on new items, repeat purchase rate and the share of revenue from products launched in the last 30 days.
Truthfulness is not optional. In the EU, the Unfair Commercial Practices Directive treats misleading claims about a product's popularity or availability as unfair practices, and in the UK the Digital Markets, Competition and Consumers Act 2024 gave the Competition and Markets Authority power to fine directly for consumer law breaches. A "Bestseller" label on a product that is not one, or a "Low stock" label driven by a timer, is a legal problem as well as a trust problem.
Measuring the badge, not the product
A fair test compares the same products with and without the badge, at the same time, for comparable visitors. The cleanest design is a holdout: badges are eligible for a set of products, and a random share of visitors (say 20 percent) never sees them.
Badge lift on badged items = (CR with badge - CR without badge) / CR without badge
Category effect = (RPV category with badges - RPV category without badges) / RPV category without badges
Shift share = 1 - (category effect / (badge lift x share of category revenue on badged items))
CR is conversion rate for visitors who viewed those products; RPV is revenue per visitor in the whole category. The first line tells you whether the badge moves the badged products. The second tells you whether the store is better off. The third estimates how much of the badge lift was demand pulled from unbadged products in the same category rather than new demand.
Here is an illustrative mini-case. A home goods store with about 300,000 monthly category visitors badges its top 20 of 220 products. Those 20 products carry 40 percent of category revenue. In a four-week holdout, conversion on badged items rises from 3.9 percent without the badge to 4.4 percent with it, a 12.8 percent lift. Category RPV rises from €1.80 to €1.84, or 2.2 percent.
Plugging in: 12.8 percent lift on items with a 40 percent revenue share would add about 5.1 percent to category revenue if none of it came from other products. The actual category effect is 2.2 percent, so roughly 57 percent of the badge lift was demand shifted from unbadged items.
The badge still helps: 2.2 percent more revenue per visitor is worth having at almost zero cost. But it is a far cry from the 60 percent "lift" the naive comparison reported, and it argues against spreading badges across another 40 products, where shift would likely grow.
Two practical limits apply. Holdouts need traffic: at 3 to 4 percent conversion, detecting a 2 percent category effect with confidence takes hundreds of thousands of visitors. Smaller stores can test one badge at a time on a high-traffic category and accept a wider margin of error. And badge tests should run at least two full weeks to cover weekday patterns.
A badge policy that survives a holdout
Once a store has measured a badge fairly, the policy usually becomes stricter, not looser. The checklist below is what tends to remain.
- Define each badge with a rule and a data source. "Bestseller: top 5 percent of the category by units sold in the last 14 days, recalculated daily." No manual badges for products marketing wants to push.
- Cap the share of badged products. If more than about one product in ten on a page carries the same badge, it stops working as a signal. Treat the cap as a starting point to test.
- Give New an expiry. 30 to 60 days for most categories, shorter for fast-fashion drops.
- Show one badge per product. Two or three badges on one card compete with the price and the photo, and the shopper reads none of them.
- Measure per audience. Bestseller on new visitors, New on returning visitors, both on category RPV.
- Keep a permanent holdout. A small share of traffic, such as 5 percent, never sees badges, so the effect can be checked again after redesigns and seasonal shifts.
- Audit the truth of every claim. Low stock and Bestseller labels must match the data at the moment they are shown.
- Watch the long tail. Track sales of unbadged products. If they fall steadily, badges may be narrowing the range customers explore, which hurts discovery and dead stock later.
The last point leads to a question no single test answers. A Bestseller badge feeds itself: badged products sell more, so they stay bestsellers, and the rest of the catalogue gets less attention every month. Over a quarter that can concentrate demand on a few products and starve new ones of the sales they need to earn a badge. Is a 2 percent lift in category revenue today worth a slowly narrowing catalogue next year, and how would you know before it happens?