The Master Conversion Rate Optimization Checklist: What to Check on Every Page, and Where to Go Deeper

Every page type needs its own checks, but twelve of them apply everywhere. Start with the universal pass below, then use the page-specific checklist for whichever template you are auditing. Working the specific list first is the common mistake — it produces detailed notes about a page whose fundamentals were never verified.

What belongs on a conversion rate optimization checklist?

A useful checklist is short enough to finish and specific enough to fail. Items like "improve messaging" cannot be answered yes or no, so they never get resolved; items like "does the headline name a specific outcome" can be checked in seconds and either pass or do not.

The twelve below apply to any page that asks a visitor to do something — a landing page, a homepage, a pricing page, a product page. They are ordered the way a visitor meets them rather than by theme, because a failure early in that sequence makes everything after it irrelevant.

Twelve is also deliberate. Lists of fifty look thorough and get abandoned around item twelve anyway, usually without anyone recording which ones were checked. A list you finish on every page beats a longer one you finish on the first.

Orientation: the first five seconds

• 1. Can a stranger say what this page offers without scrolling?
• 2. Does the page match the promise that brought the visitor here — the ad, the email, the search result?
• 3. Is there exactly one primary action, and is it obvious which one it is?
• 4. Does the largest visible element render quickly? A hero that arrives late is functionally the same as a hero that is unclear.

Persuasion: the middle of the page

• 5. Is the benefit specific enough to be falsifiable? "Save time" is not a claim; "cuts a two-hour review to ten minutes" is.
• 6. Does proof appear at the moment of doubt rather than in a block at the bottom?
• 7. Are the two or three most likely objections addressed anywhere on the page?
• 8. Does the argument survive being skim-read? Headings and opening sentences alone should carry it.

Action: the ask

• 9. Does the call to action describe what the visitor gets rather than what they do?
• 10. Is every form field one you genuinely need before delivering value?
• 11. Is the next step visible once the visitor is convinced, without scrolling back up to find it?
• 12. Does anything ask for commitment — an account, a card, a call — before the value is clear?

Which checklist should I use for my page type?

Once the twelve are clean, move to the checklist that matches the template you are auditing. Each one covers the checks that only make sense for that page type, which is why they vary in length rather than all landing on a tidy round number.

Audit templates rather than individual URLs. Most sites have three or four page patterns, and a finding on one instance almost always applies to every page using it — auditing fifty product pages individually produces volume, not insight.

How do you actually run the checklist?

Score each item pass or fail. Resist marking things out of ten — half-marks let a page look acceptable while every individual answer is vague, which is the exact state most underperforming pages are in.

Open the page at the viewport most of its traffic actually uses rather than the one you designed on, and work the items in order instead of jumping to whatever looks weakest. Twenty minutes per page is realistic for the universal twelve; a page-specific list adds another fifteen or so.

Do this on the pages that carry your traffic first. A checklist applied to a page nobody reaches produces the same satisfying list of fixes and none of the revenue, which is the most common wasted week in conversion work.

Pull your top landing pages by sessions before you start, and work down that list rather than the site menu. It takes two minutes and it is the difference between an audit that changes a number and one that changes a document.

Which failures should you fix first?

Work top-down rather than worst-first. A failure at item 2 discards visitors before they ever reach item 9, so improving a call to action on a page whose offer is unclear changes nothing you can measure.

Then weight what remains by exposure. Multiply the severity of a failure by the number of people who encounter it: a moderate problem on your highest-traffic page beats a severe one three clicks deep, almost every time. Severity alone is the wrong sort order, and it is the one most audit outputs hand you by default.

The exception is item 4. Performance problems suppress a page uniformly and are usually the cheapest thing on the list to fix, because they rarely require anyone to agree about positioning.

One more ordering rule: fix anything that loses people who had already decided before anything that loses people who never engaged. Recovering a visitor at the final step is worth more per fix than persuading one who was never close.

What a checklist cannot tell you

A checklist is a way of asking consistent questions, not a source of answers about your customers. It will tell you that a page breaks a convention. It cannot tell you whether that convention matters to the people who actually buy from you.

It also cannot establish cause. Every failed item is a hypothesis with an unknown effect size until you change it and measure the result, which means a checklist is the start of the work rather than the conclusion. If a page passes all twelve and still underperforms, the problem is upstream — traffic intent, pricing, or the offer itself — and no amount of page-level review will surface it.

Where you need to know why a visitor hesitated rather than what they saw, that is a research question. See where AI falls short in CRO for the same argument applied to automated audits.

Turning the checklist into a repeatable habit

The value of a fixed list is comparison. Running the same twelve checks against the same pages each quarter tells you whether the work you shipped moved anything; running a different ad-hoc review each time tells you nothing, however thorough each individual pass was.

Record the result somewhere durable — the landing page audit template exists for this — and change one thing at a time. Shipping four fixes at once makes the outcome unattributable, and the next audit will not be able to tell you which one earned the improvement.

Quarterly is a reasonable default for most sites, or after any change to an offer, a price, or a hero. The point is that the interval is fixed and the questions do not change between runs; both halves are what make the comparison mean anything.