CRO is not button colors
Conversion rate optimization is finding where people leave and fixing the biggest leak first. Start with session recordings and heatmaps, then the funnel in GA4 and Shopify: landing page to product page to cart to checkout to purchase. Look at mobile first, because that's where most of the traffic is. Within a day you'll know which step is bleeding.
The leaks are usually the same ones. The offer isn't clear above the fold. Reviews are buried below three screens of lifestyle imagery. The price has no anchor, so there's no bundle to make it feel reasonable. Shipping shows up as a surprise at checkout. The page takes four seconds to load. Rank them by how much money each is costing and fix the top one. Measure the lift to purchase rate, not to clicks. Then do the next one.
Cold traffic needs a pre-sell
Paid traffic that has never heard of you converts badly on a product page, because the product page assumes you already want the thing. A listicle or advertorial handles the objections first: seven reasons, what makes it different, who it's for, what people said. Then it hands off to the product page or straight to cart. Test it against the product page with the same ads and keep whichever wins. For most brands we run, it wins.
Build it, don't recommend it
Most CRO work dies in a deck. We build the pages ourselves on Shopify, Replo or Webflow so a test ships inside the week. One change at a time, measured against purchase rate. A lift in site conversion lowers cost per acquisition on every channel at once, which is why CRO runs alongside paid instead of after it.
AEO: getting cited in the answer
When a customer asks ChatGPT, Perplexity or Google's AI Overview what to buy, your brand is either in the answer or it doesn't exist. Answer engine optimization is the work of getting cited. It overlaps with SEO but it's a different game: fewer visits, far higher intent, faster to move, and it weighs different signals.
What the models reward, from what we've seen across accounts. Clean product data and schema, so the model knows what you sell, what it costs and what it's rated. Pages structured to be quoted: direct answers, specs in plain language, honest comparisons, real FAQs. And third-party presence, because the models trust what other people say about you more than what you say about yourself. Reddit threads, review sites, comparison articles, press.
How to run it
Write down thirty to fifty questions your buyer actually asks. Check who gets cited for each one today. Fix the entity and schema problems. Publish the pages the answers are missing: the comparison, the "best X for Y", the FAQ that answers the real question. Show up on Reddit and the review sites with something useful to say. Then track share of answers across your query set every month and iterate against whoever is beating you. Treat it as a testing program, not a one-time setup.
Why they belong together
AI answer traffic is small and extremely high intent. If that visitor lands on a page that leaks, the citation was wasted. And the same things that get a page cited (clear specs, real reviews, straight comparisons) are the things that make it convert. One team should own both, because they're two halves of the same job: the customer asked a question, and the answer should end in an order.
Where to start
- Pull recordings and funnel drop-off for the last thirty days, mobile first
- Fix the single most expensive leak and measure purchase rate
- Build one pre-sell page and test it against the product page
- List fifty buyer questions and check who's cited today
- Fix schema and product data
- Publish the comparison and FAQ pages the answers are missing
- Track share of answers monthly
Greg Disser
Co-Founder & COO
