A 0.5% improvement in revenue is monumental for a $500M brand and meaningless for a $1M brand. That’s why brands need different CRO (conversion rate optimization) strategies at different stages of growth.
This content was originally published in the No Best Practices newsletter on 04.28.2024.
The internet is stuffed full of “10 CRO hacks you have to try” content. But changing your add to cart button from black to blue is not going to move the needle if you’re doing under nine figures in annual sales.
That doesn’t mean that CRO is useless for smaller brands. Here are the tests you should focus on at $1M, $5M, $50M and $100M+ annual revenue.
CRO At $0-5M In Annual Revenue
Brands doing less than $5-10M/year in revenue are too small for traditional CRO to be effective. Why? These brands don’t have enough traffic or conversions to run more than one or two statistically significant tests per year.
For a full explainer of statistical significance, click here. The TL;DR: your “winners” are meaningless unless you can be confident that the test variant actually caused the lift in conversion rate (or your metric of choice).
The less traffic you run to the test, the bigger the lift needs to be if you want to achieve statistical significance. If your site is only getting a few hundred visits per day, your test will need to drive a 5-10% lift in CVR (or more) to hit stat sig.
Few tests accomplish this. So what do you do instead? Focus on testing your offer, messaging and positioning, especially if Meta ads are your primary customer acquisition channel.
Meta’s ad serving algorithm provides a shortcut around statistical significance: it will serve impressions to the audience and ad creative that has the best chance of responding. The best offer will drive the highest conversion rate. Meta will elevate the winners and suppress the losers.
Side note: Meta’s algo is probably not perfect/infallible here, but it’s much more cost effective than running true A/B tests against every ad you want to run. The “alpha” in stress testing the algo is minimal. Click here for more info on that.
Even if you’re not using Meta ads, you’ll still see one or two offers/positioning statements/etc rise to the top of whatever campaign you’re running if each variant is distinct.
CRO helps you squeeze extra juice out of a funnel that has achieved product market fit. A brand’s highest leverage testing at this stage focuses on identifying the funnel, not refining it.
At this size, do not hire a CRO agency or consultant unless they specialize in offer development and testing OR have a track record of working with clients of this size taking a “big swings” approach.
CRO At $5-50M In Annual Revenue
At this level of revenue you have a scalable, sustainable offer. But you still don’t have enough traffic to run many tests to stat sig, unless those tests are driving a huge lift. As you approach $30-50M in annual revenue, you may have enough traffic to run a stat sig test every 8-12 weeks.
Again, this does not mean “don’t test”. It just means “don’t test randomly”. A few focus areas:
Site speed: This is a great time to audit your site for all of the plugins, apps and hacky code dragging down your page load times. A faster site almost always means better conversion rates. Don’t forget to check any pages created in a landing page builder too!
As the business grows and you test more apps and tech, you should audit this quarterly. Make sure that apps you’re no longer using have been uninstalled and there is no legacy app code left behind after uninstalling (yes, this happens frequently).
Removing ambiguity: The best way to identify points of ambiguity is to have outsiders from several different demographics browse your site and try to check out. Boomers not involved in DTC are a great place to start.
Watch them use the site on their phone, make note of questions that come up or areas where they get stuck, and ask follow-up questions. You’ll have to dig deeper to get insights on issues like lack of product info, confusing photos, shipping options, etc. If something comes up more than once, it’s fertile ground for a test.
Depending on how much traffic you have, you can bundle several related changes into a single test. This won’t get you the perfect causative answer/result, but it will help you validate changes faster.
Basic UX Stuff: At this size, if you’re committed to growth via Meta ads, you need to commit to UX “best practices” to a certain degree (I know…). If you launched a highly branded website with custom fonts, or purposefully ugly design meant to appeal to Gen Z, you might want to walk it back.
Yes, a lot of UX-optimized websites look similar. Yes, we shouldn’t sacrifice all originality at the altar of CRO. BUT you’re too small to prioritize brand over the user at this stage unless you are driving most of your demand and awareness outside of digital channels.
To address these basics, pay a CRO agency or expert for a focused audit. Better yet, pay for two or three audits and implement only the recs that come up in all of them.
New acquisition funnels: At this stage in growth, unlocking a new acquisition funnel is almost always going to be more impactful than improving the conversion rate of your existing traffic.
If you’re only selling one product, or one product plus one or two cross-sells, now is the time to develop your second hero product. If you’re already selling multiple products, it’s time to figure out how to hero something else in the assortment OR develop a new hero.
A note on apps: Brands at this stage of growth can get distracted by Shopify apps. Yes, they get hyped on Twitter. No, they’re not going to solve strategic/growth issues with your business.
You really should be holdout testing every app you implement to see if it’s really driving incremental lift. At this level of revenue, that isn’t feasible. So my advice: don’t let apps become a deadly distraction. That wacky web widget is probably going to hurt your conversion rate more than help it.
CRO At $50-99M In Annual Revenue
At this level of annual revenue you typically have enough traffic to run at least one stat sig test per month. This is when testing more focused changes to the site experience can start to pay off.
You should start with the bottom of the conversion funnel–the checkout–and work your way up: checkout, cart slider, PDP, category page, homepage. Checkout testing can include shipping rates and service offerings and/or the entire set of incentives you use around free shipping.
At this size, you want to put some structure behind your testing program. Tests should be prioritized quarterly based on an estimate of their potential impact. Winning test variants that are pushed live should be tracked with a 10% holdout for at least three months to determine if the lift is maintained.
The learnings from all tests (successful or not) should be shared with your eCom and marketing teams, and those learnings should influence your roadmap. A lot of brands miss out on the full value of a CRO program because they don’t record test results consistently or use them to build a knowledge base.
At this revenue level you should also be running holdout tests on any consumer-facing app or software you install: email popups, live chat widgets, in-cart upsells, etc. These tests should go on the quarterly roadmap, just like everything else.
My personal opinion: this is the revenue level where it makes the most sense to bring on a traditional CRO agency. I’ve seen a lot of brands of this size mismanage CRO because the internal team doesn’t understand statistics or developing accurate estimates of impact.
A good agency will help you prioritize the right tests, educate your team on stats principles, and help you back away from shiny objects that are actually terrible ideas.
CRO At $100M+ In Annual Revenue
You’ve done it: you’re generating enough traffic and revenue to run multiple statistically significant tests at the same time. But with great power comes great responsibility–good statistical principles and good project management are critical to realizing the full value of CRO.
At this stage, you need a project manager who can work cross-functionally with every team running tests, including marketing. Without a PM you run the risk of staging overlapping tests that invalidate each others’ results.
If the email team is running an A/B test, but some random sample of that traffic receives a test treatment for a new landing page design, both tests are skunked.
Every brand structures their testing program differently. In some brands, a single team owns testing (and stats expertise) for both marketing and site experience. In other brands, each channel team is responsible for its own tests and the eCom/site experience team owns CRO.
You can opt to run CRO through an agency, but keeping that agency plugged in to other teams’ testing roadmaps becomes harder.
A huge opportunity you unlock at this level of scale is the ability to design and test distinct site experiences for traffic coming from different marketing channels.
At Tapestry, Meta dyanmic product carousels (then called DPA) were a huge source of customer acquisition for some of our brands. These ads force you to link to a product detail page. But if the viewer doesn’t like what they see on the PDP, that’s it. The default PDP experience gave these visitors four “related products” halfway down the page.
We built out a “DPA” version of the PDP where an abbreviated version of the key product info gave way to a grid of 12-16 related products. This was a “hacky” version of the concept; eventually we wanted to introduce infinite scroll and personalized product recs.
This variation drove a lift in conversion rate for our DABA traffic that was generating 4-5 figures of incremental revenue per day for our biggest site. Does a CRO test like this make sense for a $25M brand? Probably not. But at 9+ figures in revenue, channel-specific experiences can yield a huge benefit.