Sell the product, don't test the button

If you run a DTC brand and your conversion rate is lower than some benchmark you saw, someone has probably suggested a CRO program. A testing tool with a monthly fee, a roadmap, a standup. The first experiments are something like a sticky add-to-cart bar, a new button treatment, or reordering the modules on your page.

I want to talk you out of spending a lot of time or money on this, or at least out of doing it now. Not because conversion rate optimization doesn't work — it does, for huge brands with tons of traffic. But if you're under roughly $20M in DTC revenue, a testing program is one of the least impactful places you could put your attention for the next two quarters, and I think the reason why is worth walking through, because it changes how you should think about your conversion rate in the first place.

Let me use one brand throughout so this stays concrete. Say you run made up brand Aurelia, a fine jewelry brand. Your hero product is a $140 gold vermeil ring, you're doing $12M a year, and your site converts at 2.4%. Someone shows you a benchmark that says you should be at 3%. Sure, is it better to have higher conversion? Yes. Do I think CRO is the way to improve your conversion? No.

Two reasons. The first one is the actual argument. The second one is for the people who nod along and then want to test their way there anyway.

Reason one: CRO doesn't sell anything

Your conversion rate is mostly a demand metric, not a site metric.

It's easy to think of it as a site metric. It has "rate" in the name, it lives in your site analytics, and every tool that promises to improve it points at your site. But look at what happens when you break Aurelia's conversion rate out by traffic source. You'll see something like this:

• Branded search: ~9%
• Direct: ~5%
• Email: ~5%
• Organic social: ~1.5%
• Cold paid prospecting: ~1%

Same site (unless you're doing a ton of personalization, which at this size brand, is unlikely). Same ring, same photos, same button, same checkout. The conversion rate swings by nearly 9x, and nothing on the page changed between the top line and the bottom one. What changed is how sold the person already was when they got there. Someone typing "Aurelia ring" into Google has been sold already — by a friend, a creator, a gift guide, a ring they saw on somebody's hand. They're not evaluating you, they're looking for you. Someone clicking a cold ad has never heard of you and is being asked to hand $140 to a stranger.

So the thing moving your conversion rate the most is the selling that happens before anyone lands on the page — and then, on the page, whether it actually makes a case for the ring or just describes it. CRO doesn't really touch either of those. It optimizes the container. The selling is what goes in it.

The selling happens in a few places. Let me go through them on Aurelia's actual ring, because "sell better" is useless advice until you say what it means.

Is the ring worth $140 to this person in the first place? Nothing downstream survives getting this one wrong. If the ring is hard to tell apart from a $40 version two search results away, no navigation streamlining exercise is going to fix that — you have a positioning problem, and it will show up in your conversion rate, and it is not a CRO problem. This is the least fun item on the list and the one most worth being honest with yourself about before you spend a dollar on the rest.

Does the customer show up already sold? Most of your paid traffic is cold, which means the ad is doing the selling — or not. And the ad that sells isn't the one that describes the ring to everybody; it's the one that picks a specific person and talks only to her. An ad that shows the ring spinning on a white background with "14k gold vermeil, now available" is aimed at no one, so it lands on no one. Compare three versions that each pick a person:

• the Christmas gift that doesn't look like you asked a coworker what to buy — for the guy panic-shopping in December
• you don't need a reason, or a partner, to own real gold — for the woman buying for herself
• for the ring that lives on your hand, not in the box for special occasions — for the person who wants everyday gold, not an heirloom she's afraid to wear

Same ring, three different people, and each one feels like the ad was made for her (or him) — because it was.

Do that well and two things happen. The platform shows each ad to more of whoever responds to it, so sharper targeting finds you more of the right people. And those people land on the page already half-sold — which is why, back in that table, email converts around 5% and cold prospecting around 1% on the identical site. Same ring, same page, same button; the email list just already knows you. Your conversion rate is largely a measure of how well you sold people before they arrived.

Does the page romance the product? This one is on the site — but it's more merchandising, not CRO. Go read one of your product pages out loud. There's a good chance it says something like this, because someone pasted it in from the tech pack in 2022 and it has known sweet, sweet peace ever since:

14k gold vermeil over .925 sterling silver. 2mm band. Sizes 5–9.

Here's the same ring, sold:

Real gold that lives on your hand — through showers, workouts, and a sink full of dishes. Vermeil this thick won't turn your finger green or wear thin in a year, the way plated rings do. Most people who buy one are still wearing it daily a year (or more) later.

Same ring. Same facts, even. One answers a question about materials; the other answers the question the customer is actually asking, which is am I going to regret spending $140 on this.

Turn every spec into the reason it's there. "14k gold vermeil" is a fact. "Won't turn your finger green, and costs a fifth of what solid gold does" is a reason. Your key features row should be four reasons to buy, not four attributes off the spec sheet.

Make your photos prove something. Most brands have plenty of lifestyle shots and a wall of customer photos, but they're decorative — a beautifully styled hand that could be selling any ring. The shot that earns its place shows the claim: the ring still on in the shower, stacked three-deep on a real hand, a year of everyday wear on it. Your customers are already posting exactly these — pull the best ones onto the page and tell the customer why they should care.

Let your reviews write the copy. Go read fifty reviews of the ring. The same five reasons will come back over and over — doesn't turn my finger green, wore it every day for a year, looks more expensive than it was, stacks with everything, fit was right the first time. That's your copy. Your customers wrote it for you, for free, in their own words, and you left it in a widget three-quarters of the way down the page. Make sure these concepts are showing up above the fold, too.

Does the page answer the fear? For a $140 ring someone's buying without trying on, the worries are real and specific: will it fit, will it turn green, what if I hate it?! Your return policy and your warranty are some of the strongest reassurance you've got, and right now they're a gray link in the footer that reads like a legal disclaimer. Say the good version, on the page, in plain words — free resizing, lifetime tarnish warranty, ships tomorrow. That does more to close the sale than any badge you could add.

What all of these have in common is that they're the selling — the product, the ad, the story on the page, the reassurance. CRO isn't really selling. It's tidying up after the sale is mostly already won or lost, and at Aurelia's size, it's won or lost upstream. That's where your attention belongs.

Reason two: even if you wanted to test, you can't get a clean read

Maybe none of that moved you — you like rigor (I get it, I do too), and you'd rather test the copy and the photos and the badges and let the data decide, rather than trust your gut on what sells. The problem is you don't have the traffic.

Let's say Aurelia does 2.4% on about 300,000 sessions a month. Here's how long it takes to reliably detect a given improvement:

• A 5% lift (2.40% → 2.52%): ~8 weeks
• A 3% lift (2.40% → 2.47%): ~5 months
• A 2% lift (2.40% → 2.45%): ~11 months

A button or badge test, if it does anything at all, lives in that bottom range. So picture the best case on the 2% row: eleven months to confirm a win, during which you can run no other test, hold no sale, launch nothing, and hope your traffic mix doesn't shift — all to move from 2.40% to 2.45%. Your testing program's ideal year is one where your marketing team ships nothing, sells nothing, and changes nothing on the site. If you've got a marketing team that's actually doing that, please notify the proper authorities for a wellness check.

And that's the good case — it assumes you're testing the whole site at once. Test a single product page and it's hopeless; your hero ring might see 60,000 sessions a year, and everything below it far less. There's no version of this math where a page-level test teaches you anything.

Here's the part that closes the door for me. A test is for settling a question you can't call by eye — nobody A/B tests whether to fix a checkout that's throwing errors, you just fix it. You reach for a test when a change is too subtle to judge. But subtle changes move the number by a little, and a little is exactly what your traffic can't measure. So the big changes don't need a test, and the small ones can't get one. Under $20M, there's not much room in between.

So, sell the ring

Make sure the ring is worth $140 to somebody. Get it in front of the right people, through channels where a stranger can be turned into a believer. Make your ads argue instead of announce. Then make the page actually romance the product — sell the ring instead of describing it, let the photos and the reviews carry the weight, and answer the obvious fears before they're asked. Do all of this because it's plainly better, not because a test signed off on it. My rule of thumb: if you need a test to tell you whether a change is an improvement, the change is probably too small to matter at your size.

Then later — once you're big enough that a 2% lift is worth eleven months of someone's attention — go ahead and test the button. It'll still be there. It's just that it's sand, and you've got boulders to move first.

Looking at your conversion rate and not sure where the real lever is?

Nine times out of ten it's upstream of the page — in the product, the offer, the traffic mix, or the story you're telling on the page. If you'd like a second set of eyes on which one it is for your brand, send me a note.