How to Prioritize CRO Fixes When Everything Looks Broken
Somebody looks at your site, or you finally look at it properly yourself, and now you have a list. Nineteen things. All of them defensible. None of them obviously first.
So the list sits there. This is the most common way conversion work dies, and it isn't a motivation problem. It's that an unranked list is genuinely paralyzing, because every item on it looks equally like work.
Here's how I rank findings, and why the popular frameworks tend not to help at small scale.
Is your site converting as well as it could be?
Get a CRO Audit - $99Why ICE and PIE probably aren't your tool
If you've read about prioritization you've met ICE (impact, confidence, ease) or PIE (potential, importance, ease). Both are fine, and both were built for something you're probably not doing.
They're designed to rank test hypotheses for a team running a continuous experimentation program. The scoring assumes you'll validate each one, which means being wrong is cheap and you're really deciding what to test next.
At small scale you aren't testing. You don't have the traffic to. You're deciding what to change, once, and living with it. That's a different question, and it needs a harder look at consequence than a 1-to-10 ease score gives you.
Start by separating blockers from improvements
This is the distinction that does most of the work, and almost nobody makes it.
A blocker is something that makes conversion impossible, not unlikely. A broken checkout. A contact form that doesn't send. A payment method that errors. A page that doesn't load on mobile.
An improvement makes conversion more likely. A better headline, stronger trust signals, a clearer CTA.
These are not two ends of one scale. They're different categories, and every blocker outranks every improvement, no matter how good the improvement is.
I audited a site recently where the contact link was malformed on every page. It displayed correctly, so it looked fine, but clicking it opened a message to an invalid address. There were no forms anywhere on the site. That owner could have rewritten every word on his homepage and gained nothing, because there was no way to act on being persuaded.
Find your blockers first. Test every path to money on your own site, on your phone, today. Most sites have none, some have one, and the ones that have one usually don't know.
Then rank improvements on four questions
Once blockers are handled, I score everything else by asking four things in this order.
1. How many people hit it? A problem on your homepage affects everyone. A problem on a product page affects the people who reach that page. A problem in your checkout affects only people already trying to buy, which is fewer people but a much more valuable group. Weight by traffic, not by how much the problem annoys you.
2. How close is it to the money? Friction gets more expensive the further down the funnel it sits, because the visitor has already invested attention and you've already paid to get them there. A confusing headline loses a cheap visitor. A confusing checkout loses one you've already spent everything to acquire.
3. How sure am I that fixing it helps? Be honest here, because this is where opinion sneaks in. "The submit button doesn't work" is certainty. "This headline would convert better" is a hypothesis. Both can be worth doing. Only one deserves to jump the queue.
4. What does it cost to fix? Last, deliberately. Cost is a tiebreaker, not a driver. It belongs at the end because putting it first is how people end up doing seven trivial things and none of the important ones.
The question that settles most ties
When two items score similarly, I ask the inverse: what happens if I never fix this?
It's a more useful question than "how much will this help," because it's easier to answer honestly. "If I never publish my turnaround time, people with deadlines keep quietly choosing someone who did" is concrete. "This might lift conversions 8%" is a number you made up.
Most of the time one item has a real answer and the other has a vague one. That's your order.
Three ways people get this wrong
Doing the easy ones first. The most common, because it feels productive. You clear six small items in an afternoon, the list looks shorter, and none of them touched revenue. Easy work is still work, and a Saturday spent on favicons is a Saturday not spent on your checkout.
Doing the visible ones first. Design problems are the ones you notice every time you open your own site, so they nag. But the things nagging you are the things you've looked at a hundred times, which is exactly the opposite of your visitor's experience. Your irritation is not a ranking signal.
Redesigning instead of fixing. When the list is long, a rebuild feels cleaner than working through it. It almost never is. A redesign takes months, costs real money, changes a hundred variables at once so you learn nothing, and routinely reintroduces the same problems in new clothes. If your list has fifteen items, fix four. Do not build a new site.
Do this before any of it
Install analytics first if you don't have it. GA4 and Microsoft Clarity are both free and take about fifteen minutes together.
This feels like a delay and it isn't. Without it you can't answer question one, you have no idea which pages people actually land on, and you'll never know whether anything you did worked. A week of session recordings will also reorder your list on its own, because watching ten people struggle with something you thought was fine is more persuasive than any framework.
What good prioritization looks like
A finished ranking is short at the top. Blockers, then three or four improvements that are high-traffic, close to the money, and high-certainty. Everything else goes in a "later" pile you're allowed to ignore.
That last part matters. The point of ranking is not to schedule all nineteen items. It's to give you permission to not do fifteen of them right now.
This is most of what you're buying in an audit. Finding problems is the easy half, and a free scanner will hand you a list. Knowing which four are worth your week is the part that requires looking at your specific business and making a call.
If you want that call made for you, here's a real audit with its findings ranked, and the audit itself is $99. If you'd rather do it yourself, start with the checklist and rank what it finds using the four questions above.