A CRO agency is the team you hire to get more sales out of the visitors your store already has. That is the definition, and it is the easy part. The more useful questions are what one actually does week to week, what it costs, and whether your store is ready for one yet. We run a CRO agency for Shopify and WooCommerce brands, so bear that in mind as you read. You will also find a section below on when we would tell you not to hire one.
| The short answerA CRO (conversion rate optimisation) agency finds out why visitors are not buying, fixes the causes, and proves which fixes worked. The work runs in a loop: research, prioritise, test or build, measure, repeat. A good one judges itself on revenue per visitor rather than clicks or opinions, and builds the winning changes into your site rather than handing you a list. It is worth hiring once you have enough traffic to test properly and the store is already spending money to attract visitors. Below that, most of the same value comes from a one off audit. |
What a CRO agency actually does
Strip away the jargon and the work falls into six jobs, in roughly this order.
1. Fix the measurement first. Nearly every store we audit has tracking problems: purchases recorded twice, add to cart events missing on some templates, consent settings hiding a large share of sessions. Every decision after this point depends on the numbers, so a CRO agency that skips this step is testing on sand.
2. Research why people are not buying. Two kinds of evidence. The numbers show where visitors drop out: which templates, which devices, which traffic sources. Session recordings, heatmaps, on site surveys and a heuristic review of every template show why. The output is a list of specific problems, each with evidence attached.
3. Prioritise the backlog. Not every problem is worth fixing first. Each one is ranked on how many visitors it affects, how strong the evidence is, and how cheap the fix is to build. A delivery cost surprise at checkout that affects every buyer beats a homepage banner that almost nobody scrolls to.
4. Test, or just build. Where there is enough traffic, the change is run as an A/B test against the current version. Where there is not, or where the fix is plainly a bug, it is built and measured before and after. Knowing which is which is most of the skill.
5. Ship the winners properly. A winning test left running inside a testing tool’s script slows the site and breaks at the next theme update. The winner should be built into the theme itself. Many CRO agencies stop at step four and hand this back to you.
6. Report and repeat. Every test is written up, win, lose or flat, with what it taught you about your customers. Losing tests are not wasted. They tell you what your customers do not care about, which stops you spending money on it again.
What a CRO programme looks like, month by month
| When | What happens | What you should have at the end |
|---|---|---|
| Weeks 1 to 4 | Tracking audit and fixes, analytics review, recordings and heatmaps, template by template review | Measurement you can trust and a prioritised backlog, each item with evidence and an estimated value |
| Month 2 | Quick fixes built straight away, first tests launched on the highest value items | Bugs gone, first tests running for whole weeks |
| Month 3 | First results called, winners built into the theme, next round launched | Your first measured changes to revenue per visitor |
| Month 4 onwards | A steady cadence of tests, each informed by the last | A growing record of what your customers respond to, and gains that stack |
If an agency promises results in the first fortnight, ask what they will be measuring. Research and tracking fixes can move things quickly. Properly run tests cannot, for reasons the next section makes very concrete.
Does your store have enough traffic to test?
This is the question most CRO agencies avoid, because the answer rules out a lot of potential clients. An A/B test needs a certain number of visitors before the result means anything, and that number is larger than most people expect.
Take a store converting at 2%. To reliably detect whether a change lifts that to 2.2% (a 10% relative improvement), you need around 80,700 visitors in each version, so about 161,000 in total. To detect a bigger lift, 2% to 2.4%, you need around 42,000 in total. Those figures use the standard two proportion calculation at 95% significance and 80% power, which is the convention most testing tools default to. You can check your own numbers with a free sample size calculator.
Here is what that means in time, if every visitor to the tested pages is split evenly between the two versions:
| Monthly visitors to the page being tested | To detect a 10% lift | To detect a 20% lift |
|---|---|---|
| 10,000 | About 70 weeks | About 19 weeks |
| 25,000 | About 28 weeks | About 8 weeks |
| 50,000 | About 14 weeks | About 4 weeks |
| 100,000 | About 7 weeks | About 2 weeks |
Based on a 2% baseline conversion rate. Stores converting higher need fewer visitors, and tests measured on a more frequent action, such as add to cart, need fewer again, which is why good programmes often test further up the funnel first.
Two things follow from this. First, at lower traffic levels you can only test big, bold changes, because small tweaks would take more than a year to read. Second, any agency calling a test after a few days on a store of this size is reporting noise. The number that looks like a win on a Tuesday is very often gone by the following Monday.
If you are not sure where your conversion rate should be to begin with, our guide to what a good eCommerce conversion rate is covers the benchmarks and why they vary so much by sector.
What to do if you do not have the traffic yet
You still need conversion work. You just need a different shape of it.
Below the point where split testing is practical, most of the value sits in steps one to three above: fix the tracking, find the obvious leaks in the recordings and analytics, and build the fixes. Broken filters, slow product pages, delivery costs that only appear at checkout and a mobile add to cart button below the fold do not need a test to prove they are problems. Fix them, measure before and after, and move on.
That is usually a fixed fee audit rather than a monthly retainer, and it is often the better buy. Our free Shopify store audit is a quick way to see the first things worth fixing before you spend anything. And if the real problem is that not enough people are arriving in the first place, conversion is not the lever to pull first. Our guide on how to increase eCommerce sales sets out how traffic, conversion, order value and repeat purchase fit together.
CRO agency or the alternatives?
| Option | Good for | Watch out for |
|---|---|---|
| A testing tool on its own | Teams that already know what to test and how to read results | The tool runs tests. It does not tell you which ones are worth running |
| A freelance CRO consultant | Research and a strong backlog at a lower cost | Someone still has to design and build every test and every winner |
| An in-house CRO hire | Larger stores with enough traffic to keep someone busy all year | One person rarely covers research, UX, analytics, design and development |
| Your general marketing agency | Joining conversion up with the traffic they already run | CRO often becomes a line in the proposal rather than a programme with its own rigour |
| A specialist CRO agency | Stores with real traffic that want research, testing and build done properly | Check that they build the winners, not just recommend them |
How to choose a CRO agency
Six questions sort the good ones from the rest quickly.
- What is your primary metric? The right answer is revenue per visitor, with conversion rate, order value and margin watched alongside it. A test that lifts conversion by discounting everything is not a win.
- How do you decide how long a test runs? You want to hear that sample size is calculated before launch and tests run for whole weeks. If the answer is “until one version is clearly ahead”, walk away.
- Who builds the winning changes? If the answer is your developers, budget for that time, or choose an agency that builds on your platform.
- What happens to our site speed? Testing scripts add weight. Ask how they are loaded, and whether they are removed once a test ends.
- Can we see write ups of losing tests? An agency that only shows winners is showing you marketing, not a programme.
- Who does the work? Find out whether the senior person in the pitch is the one running your account.
And two red flags. Be wary of guaranteed uplift figures before anyone has seen your data, because nobody can know. Be equally wary of headline “win rates”, since a high win rate usually means the tests were too timid or the results were called too early.
CRO on Shopify is its own job
On Shopify, a lot of conversion work happens in places a generic CRO agency does not touch: product page templates, collection filters and sort order, app scripts slowing the store down, and, on Shopify Plus, checkout extensions. Tests are usually run with Shopify native testing apps such as Shoplift or Intelligems, and winners need building into the theme’s Liquid templates rather than left running as overlays. It rewards a team that builds Shopify stores as well as testing them. That is how our Shopify CRO programmes are set up, with the same team researching, testing and building. If you want the whole picture of what we do on the platform, see our Shopify agency page.
Frequently asked questions
Conversion rate optimisation. Your conversion rate is the share of visitors who complete the action you care about, usually a purchase for an eCommerce store. Optimising it means increasing that share without needing more traffic.
A/B testing is one tool inside CRO. CRO is the whole process: tracking, research, prioritising, then testing where traffic allows and simply building where it does not. An agency that only runs tests skips the part that tells you what to test.
Pricing usually comes in one of three shapes: a fixed fee audit and research sprint, an ongoing monthly programme, or occasionally a fee linked to performance. What drives the cost is how many tests run each month and whether the agency builds the changes or leaves that to you. For reference, our ongoing CRO programmes start from £4,000 a month on a rolling basis, and a standalone audit is a fixed fee over three to four weeks.
It is worth it when a small lift in conversion is worth more than the fee. Work it out from your own numbers: monthly revenue multiplied by the lift you could realistically find. If you are already paying for traffic through Google Ads, Shopping or paid social, every point of conversion improvement makes all of that spend work harder. If traffic is low, a one off audit usually gives better value than a retainer.
Tracking fixes and obvious leaks can show up within weeks. Properly run tests usually produce their first reliable results in the second or third month, depending on traffic. Gains from a testing programme build over months as winners stack.
It can if tests remove content or internal links that pages rank on, or if testing scripts slow the site. Both are avoidable. Ask your agency to measure page speed before and after changes and to check template changes against what each page ranks for.
A tracking check, an analytics review of where visitors drop out, session recordings and heatmaps, a review of every key template, and a prioritised list of problems with the evidence and estimated value of fixing each one. A good audit is useful on its own, whether or not you go on to run tests.
Revenue per visitor first, then conversion rate, average order value and margin. Each test should come with a write up of the evidence, the hypothesis, the result and what it taught you, including the tests that lost.
Related reading: What is a good eCommerce conversion rate? and How to increase eCommerce sales.
| Wondering whether your store is ready for CRO?A 30 minute call with the co-founders who run our CRO programmes. We will look at your traffic and tell you straight whether a testing programme, a one off audit or neither is the right next step. Book a growth call. |