What converts online changes depending on what you sell

For an equivalent product, with the same checkout and the same technical care, a fine jewelry site converts at around 0.7% while a food site clears 5%. Dynamic Yield's e-commerce benchmarks (opens in a new tab) (global rolling twelve-month average, pulled in June 2026) show 5.29% for food and beverage, 4.71% for beauty, 2.81% for fashion, 1.2% for home and furniture and 0.7% for luxury and jewelry. Seven times less, for comparable interface work.

That gap does not measure the quality of the sites. A jewelry site at 0.7% can be flawless and a food site at 5% perfectly ordinary. It measures the nature of what is being sold. You do not order a ring the way you reorder your pasta and a site that ignores this difference applies the right levers at the wrong moment.

The good news is that this nature is not mysterious. A site's conversion ceiling can be read from three observable variables, measurable before a single line of code is written. They are what decide what your site has to do in order to sell.

Place any product on three sliders and you get a reliable prediction of how it will convert.

Dynamic Yield's gap then unfolds with no mystery. Food stacks the three favorable factors: low basket, no deliberation, constant repurchase. Luxury stacks the three unfavorable ones: very high basket, long deliberation, rare repurchase. Everything in between sits along that gradient. The rate is not a quality score, it is the signature of a category.

Worldwide average e-commerce conversion rates over twelve months (source: Dynamic Yield, pulled in June 2026). All-sector global average: 2.66%.

Behind these three variables lies a distinction marketing research drew long ago and that has not moved since. In 2000, Ravi Dhar and Klaus Wertenbroch published in the Journal of Marketing Research a study (opens in a new tab) that became a reference. They separate two families of products. Utilitarian goods serve a practical goal and are judged on efficiency: detergent, batteries, a software subscription. Hedonic goods belong to experience, pleasure and desire: perfume, jewelry, a designer piece. You need one, you want the other.

One point from that research lineage changes everything for a site. In 2005, in the same journal, Erica Mina Okada showed that a hedonic purchase needs to be justified (opens in a new tab) far more than a utilitarian one. Nobody justifies buying toothpaste. Buying a 3,000-euro piece of jewelry, they do. That is exactly what a good site for a desirable product does: it gives the visitor reasons to give in to the urge, through story, material and staging. The utilitarian product needs no narrative. It needs to be left alone.

Sites are often split between the showcase, which would display, and the conversion site, which would sell. The distinction is convenient but it does not hold. A showcase site that never triggers a call or a quote request has simply failed at its job. Every site seeks to convert, whether through a payment, a form or a phone call. The real question is not whether yours should convert but understanding what, in your visitor, triggers the decision. And that depends entirely on what you sell. Here are the four logics that cover most cases.

Food, hygiene, cleaning, consumables. The visitor already knows what they want on arrival. There is no desire to create, only a practical decision not to spoil. That is exactly why food posts the highest rate of any sector at 5.29%: clear need, low basket, frequent repurchase. Everything plays out after the "add to cart" click, inside the checkout.

And that is where sites lose the most. Drawing on fifty aggregated studies (recorded in September 2025), the Baymard Institute (opens in a new tab) documents an average cart abandonment rate of 70.22%. The leading avoidable cause is clear: 39% of shoppers give up because of unexpected costs discovered at payment. Next come delivery judged too slow (21%) then forced account creation (19%). On the form side, an average checkout shows close to 24 fields where 12 to 14 would do. Baymard estimates a better checkout can return up to 35% of conversion.

The trap: over-styling at the expense of legibility. A site for everyday products that hides its price behind a polished staging frustrates more than it seduces. On the utilitarian, design is precious when it speeds the decision, never when it slows it. That is the whole point of a CTA that converts (opens in a new tab), clear and direct.

Beauty, fashion, decoration, small designer objects. The visitor has no urgent need, the desire has to be created. Yet on a screen, everything they perceive of the product is what the interface shows them. The quality of the experience becomes, quite literally, the perceived quality of the product. Beauty and fashion convert respectably (4.71% and 2.81%) because the basket stays affordable, but the sale is not won at the checkout: it is won earlier, on the product page.

This lines up with what research says. If a hedonic purchase needs to be justified, then the site's role is to supply that justification and it runs first through the senses. The image is not a cost line to compress, it is the engine of the sale. A photograph that renders the material, the light and the true scale of a product does two things at once: it triggers the desire and it cuts the doubt, which brings returns down.

The trap: treating a pleasure brand like a functional catalog. A grid of tight thumbnails with a struck-through price everywhere turns a desirable object into plain commodity. You do not discount desire, you build it.

Luxury earns its own category because it inverts the rules. Its 0.7% rate is the lowest of any sector and that is not a failure, it is the model. Where classic e-commerce hunts down the slightest friction, luxury adds it on purpose.

On a fine jewelry or fine watchmaking site, you rarely find an instant buy button, often no price shown, sometimes just a "book an appointment". This restraint is not an oversight, it is a filter. It signals that the object cannot be grabbed in one click and it selects serious intent. Research lights up why it works: if a hedonic purchase is built and justified, a slow, staged journey is not an optimization flaw, it is what the client comes for. Forcing an express checkout with a countdown on them would destroy the value they came to buy.

The misreading to avoid: deliberate friction does not mean a slow or confusing site. A luxury site has to be technically flawless. The slowness is in the narrative pace, never in the load time. A prestigious site that genuinely lags destroys the very perception it is trying to build.

Lawyer, architect, agency, consultancy, craftsman. Here the visitor is not buying an object, they are committing to a person or a team and the decision hangs on one question: can I trust them? Conversion is rarely an instant payment, it is a first contact (call, form, appointment). The site's role is to lower perceived risk before that first step.

The Nielsen Norman Group has studied perceived site credibility since 1999. Four factors (opens in a new tab) recur, stable over twenty-five years and across every device: design quality, upfront transparency, complete and current content and connection to the rest of the web (presence, mentions, external proof). On a service, these signals do not decorate the page, they are the product.

The trap: rushing. False urgency ("only 2 spots left") erodes a service buyer's trust instead of creating it. The decision often matures off the site, in a conversation or a recommendation. The site sets the stage, it does not force the hand.

Four conversion logics, readable along the need / desire axis.

Few businesses fit a single box. A craft baker sells the utilitarian (the daily bread) and the hedonic (the celebration piece). A cosmetics brand blends desire and repurchase. So the right question is not "which box am I in?" but "what triggers the decision for THIS product on THIS page?". Three questions are enough to settle it.

The answers dictate the interface and not the other way around. And the right performance marker is never the worldwide 2.66% average: it is your own sector's benchmark. A jewelry site at 1% is overperforming when a food site at the same rate is in trouble. Measuring your conversion without knowing your family means aiming at a number that means nothing. That is the diagnostic that extends, at the product level, what was already visible at the sector level in what converts a SaaS does not convert an agency (opens in a new tab).

Copying the site that converts in another sector is the surest way to sabotage your own conversion. The ultra-optimized checkout that works wonders on a consumable can scare off a jewelry client and the staging that elevates a ring can make someone who just wants to reorder their detergent lose patience.

Before thinking colors, buttons or animations, only one question truly matters: does what I sell belong to need or to desire, and what does that change for the person deciding? Everything else follows. A site that converts does not apply the right tactics in the abstract. It applies the tactics that are right for what it sells and turns that fit into an experience the visitor feels without even knowing how to name it.