Education
The eight things we check first.
Every audit starts somewhere. These are the eight spots most likely to be quietly losing you sales. Flip a card to see what good looks like, and what usually breaks.
Say what you sell in five seconds flat.
A homepage is a storefront window, not a brand museum. In one glance it should say what the store is, show the best product, and give someone an obvious place to click.
What works
- Hero image features the best-selling product in use, on a darker background so white text pops, not a clean studio shot sitting in isolation.
- Hero copy does three jobs at once: what the store is, what the best product is, and proof that real people like the brand.
- Hero description stays at two lines max. If someone can't get the offer in the time it takes to walk past a window, the hero failed.
- A dedicated Browse Best Sellers section that auto-updates from actual top sellers, not a curated guess.
- Homepage-to-product paths rebuilt around where dead-click and scroll-depth data show real engagement, not where the owner assumes it.
- Real customer UGC near the footer, updated regularly. Stale UGC reads as fake.
- A rotating promo bar for shipping thresholds, sales and gift-with-purchase. Cheap, high-leverage real estate.
- Hero copy tested like a headline test: "Protected, for every ride." / "Gear That Has Your Back." / "Stay Protected. Look the Part." all shipped as A/B variants instead of picked by committee.
What doesn't
- Complex multi-product bundles as the first-time-buyer entry point. Simplify to an approachable starter offer.
- Trying to introduce the full catalog above the fold. Bury everything that isn't a best seller.
- A hero with no clear CTA path. Shop Now beats a vague tagline with nothing to click.
- Treating the homepage as a brand museum instead of a sales floor.
Answer every question before they have to ask it.
A good product page makes buying online easier than buying in a store. Details, sizes, materials, shipping, returns, all in one place, with as little friction as possible on the way to checkout.
What works
- Minimum 5 to 7 photos per product: multiple angles, close-ups, lifestyle context, and at least one clean studio shot.
- At least one video showing the product in actual use.
- Star rating and review count directly above or beside Add to Cart, not on a separate reviews page.
- Three or more specific testimonials per product that answer real objections: fit, quality, shipping.
- Long descriptions broken into scannable sections or accordions: protective elements, comfort, certification, care, shipping.
- Product technology and certifications shown in the image gallery itself, not just described below the fold.
- Bundles as their own product type, offered below the core product, not forced in as a variant.
- Volume-based quantity selectors that make buying more the obvious choice.
- Custom FAQ accordions per product instead of one generic site-wide FAQ.
- Reviews scoped to the exact product version being viewed. We caught a 3.0 product pulling reviews from the 2.0 and from a bundle, actively confusing buyers.
What doesn't
- Discount toggles that feel like a trick, like "toggle off protection to save $20." Bake the value into the price instead.
- Leading with the gift-with-purchase instead of the product. The GWP is a line item near Add to Cart, not the headline.
- Long scroll depth with certifications and trust content buried far down the page.
- Letting reviews mix across product versions or bundles.
Make the threshold visible or it won't work.
Promotions work when they're relevant and visible. They stop working the moment they become a calendar obligation.
What works
- Tiered, visible progress bars toward free shipping or a GWP threshold in the cart. People respond to seeing how close they are.
- Gift-with-purchase tied tightly to relevance, like a themed free add-on during a themed promo, rather than a generic discount.
- Testing real threshold numbers instead of guessing. A live example: free shipping at $200 vs $250, 70/30 traffic split, run through the cart platform.
- Concentrating promo effort on the two calendar moments that actually move revenue for your business.
What doesn't
- Bundling products to a holiday just because the calendar says so. It rarely works unless the product is obviously aligned with the occasion.
- Treating every micro-holiday as a promo opportunity. Chasing every date is like chasing your tail.
- Selling at customers instead of informing them. People respond better to understanding an experience than to being pitched a discount.
- Discount-stacking on already high-margin bundles instead of building discount logic around volume or loyalty.
Where most businesses actually lose the sale.
Cart and checkout should hold zero surprises: the final number, the delivery date and the return terms all visible before the last click. It's also the best place for smart upsells.
What works
- Treat cart upsells as merchandising strategy, not just another app. The goals: raise first-item cart value and help people cross a free-gift threshold naturally.
- Bundles and two-to-three item packs consistently outperform single-item upsell pushes.
- One-size-fits-all products like hats and accessories make the best low-friction cart add-ons. No variant decision required to say yes.
- A visible, working remove option on any add-on line item, and once removed it stays removed. A real bug we found: an add-on that kept re-adding itself.
- Watching dead clicks and rage clicks in checkout as a diagnostic. One audit found 7% of users dead-clicking in checkout before a word of copy changed.
- A duplicate or preview cart so customers can sanity-check the order before committing.
What doesn't
- Add-ons that behave unpredictably, or pricing that doesn't match between cart and checkout. Trust is highest-stakes right at payment.
- Single-item upsell prompts with no bundling logic behind them.
- A cart with no visual progress toward an incentive. Invisible thresholds convert far worse than visible ones.
Name them after intent, not internal taxonomy.
Collections should feel curated. Every mature store has a handful of forgotten collections quietly working against clean navigation.
What works
- Name collections around customer intent. "Grilling Essentials" beats "Sauces." "Beginner Kit" beats a plain product-type label.
- Keep each collection to roughly four to eight products. Enough to feel curated, not another catalog dump.
- Run an analytics pass per collection (sessions, conversion rate, bounce, session duration) before deciding what to keep, merge or archive.
- Clean, bold, legible tile text with the product name and one or two keywords visible on the tile itself.
- Know the platform tradeoff first: smart collections auto-populate from rules, manual collections are static and simpler for curated sets. This can't be changed after creation.
What doesn't
- Collections that made sense once and were never revisited.
- Cramming collections into primary nav just in case.
- Sparse, low-quality tiles with no clear product name or hook.
Match the page to the promise that got the click.
Every traffic source deserves a page that continues the conversation it started. A generic collection page can't do that.
What works
- Build a dedicated landing page per campaign and traffic source instead of sending every ad or email click to the homepage.
- Coordinate with whoever owns email and paid social so the page exists before the campaign launches.
- Let heatmaps and scroll depth decide what goes above the fold, not assumption.
- Use landing pages as the home for educational content a collection page can't hold: buying guides, comparisons, how-to-choose breakdowns. Exactly what this page is doing.
What doesn't
- Sending email, paid social and organic to one generic page and expecting the same message to convert all of them.
- Building the landing page after the campaign is already live.
The ad and the site should tell the same story.
Disconnected messaging between an ad and the page it lands on is one of the cheapest problems to fix and one of the most commonly ignored.
What works
- Custom landing pages built in direct coordination with email and paid social so the page feels like a continuation of the ad.
- Matching site-side incentives, like free-shipping thresholds and GWP tiers, to exactly what outbound marketing promises.
- Pausing spend to any page that no longer matches reality. Export out-of-stock products and cross-check them against live ad URLs.
What doesn't
- Running paid traffic to pages that haven't been checked for broken links, sold-out products or mismatched offers. A site crawl belongs in the pre-campaign checklist.
- Disconnected messaging between what an ad promises and what greets someone on-site.
The philosophy underneath all of it
Test first, perfect later.
Test first, perfect later
CRO is directional. It isn't about shipping a perfect page on the first try.
Qualitative before quantitative
Below roughly 100 orders or 10k sessions, a test isn't statistically meaningful. In that window, watching real behavior matters more than a dashboard number.
Session recordings detect UX issues
They're for spotting people struggling with navigation, clicking dead elements or looping. They are not a KPI tool.
Use both data types together
A/B testing tells you what happened. Behavior and session data tell you why.
Remove friction first
Add complexity back only if data proves it helps. Most stores fail on speed, clarity and basic UX long before they need anything sophisticated.
CRO decisions are reversible
Paralysis waiting for the right answer costs more than shipping a reasonable test and adjusting.
Not sure which of these is your problem? The quiz takes about a minute (and no, you can't fail it).