If you're looking for a bag of optimization tactics, wrong post. That's over in conversion rate optimization tactics. This is the audit: how to find out which part of your SaaS funnel is actually leaking, before you change anything.
I analyzed 521 SaaS landing pages building pages.report. The most common failure isn't a bad page. It's founders auditing the wrong stage of the funnel entirely.
TL;DR
- A SaaS CRO audit reviews your landing page and signup flow against stage-by-stage benchmarks, not one blended number.
- SaaS landing pages convert at a 3.8% median, well under the 6.6% all-industry median (Unbounce, 41K+ pages).
- Your trial model changes everything. Opt-in trials: 8.5% visitor→trial, 18.2% trial→paid. Card-required: 2.5% and 48.8% (First Page Sage).
- Both models land near the same visitor→paid rate (~1.2-1.5%, derived). So audit the stage that's below benchmark, not the page you happen to dislike.
- Five checks: hero clarity, pricing clarity, proof quality, signup friction, and CTA-to-buyer-stage match.
What is a SaaS CRO audit?
A SaaS CRO audit is a structured review of your landing page and signup flow that finds where visitors drop out before becoming trials or paying customers. It differs from a generic conversion audit in one critical way: every number gets judged against your trial model. An 8% visitor-to-trial rate is strong for an opt-in funnel and impossible for a card-required one. The output is a ranked list of leaks by funnel stage, not a list of page opinions.
That's the definition. The generic step-by-step process (analytics setup, heuristic review, prioritization) lives in the pillar guide: how to do a CRO audit. This post covers what's different when the product is SaaS.
If you just want a fast single-page pass, there's a 15-minute landing page audit walkthrough for that. Selling physical products? Different leaks entirely: see the ecommerce CRO audit.
The benchmarks: know your trial model first
Your landing page conversion number means nothing until you know which trial model you're running. This is the part almost every "is my conversion rate good?" thread on X gets wrong.
Start with the page itself. Unbounce's Conversion Benchmark Report is built on 41K+ landing pages (Q4 2024 data). It puts the SaaS landing page median at 3.8%, versus 6.6% across all industries. Within SaaS, hardware pages do best at 4.1% and data & infrastructure worst at 3.3%. So if you're comparing your SaaS page to a generic "good landing pages convert at 10%" claim, stop. You're benchmarking against ecommerce coupon pages.
Now the funnel. First Page Sage's data from 86 SaaS companies (71% B2B, agency client data from Q1 2022 through Q3 2025, updated September 2025) splits trial funnels by model:
| Funnel stage (organic traffic) | Opt-in trial (no card) | Opt-out trial (card required) |
|---|---|---|
| Visitor → trial | 8.5% | 2.5% |
| Trial → paid | 18.2% | 48.8% |
| Visitor → paid (derived) | ~1.5% | ~1.2% |
The derived row is my math, not theirs: 8.5% × 18.2% ≈ 1.5% for opt-in, and 2.5% × 48.8% ≈ 1.2% for opt-out, both computed from First Page Sage's stage inputs.
Look at what that table actually says. The two models end up within half a point of each other on visitor→paid. The difference is where the funnel leaks. Opt-in funnels bleed after the trial starts. Card-required funnels bleed on the signup form.
So here's the audit rule that follows: find the stage that's below its benchmark, and audit that. If your visitor→trial is at 8% on an opt-in funnel but trial→paid is 6%, your landing page is fine. Your onboarding is broken. Redesigning the hero won't fix it, and no landing page audit will find it.
This is also why you audit before you test. At Bing, fewer than a third of tested ideas moved their target metrics (Kohavi, Microsoft, 2013 - dated but still the canonical large-scale number). Testing random ideas against the wrong funnel stage compounds both problems.
The five SaaS landing page checks
Once you know which stage you're auditing, run these in order. They're sequenced by how much damage each one does.
1. Hero: the 5-second stranger test
Show your hero to someone outside your company for 5 seconds. Then ask two questions: what does it do, and who is it for? Miss either one and nothing below the fold matters.
Across the 521 pages I analyzed, this is the highest-frequency failure. Not ugly design. Heroes that read "the modern platform for teams" and could belong to forty other products. If you want to see what passing looks like, the SaaS landing page examples post walks through real heroes section by section.
2. Pricing page clarity
Plan confusion kills trials before the signup form ever loads. Check three things: can a visitor tell which plan fits them in under 30 seconds, is the real price visible (not "starting at"), and is the trial's relationship to the plans explicit? "Free trial, then what?" is a question your pricing page must answer on the page.
If your ACV doesn't justify "Contact sales," show the number. Visitors who can't see the price assume the worst and leave.
3. Proof that survives a skeptic
SaaS buyers in 2026 are numb to logo walls. A row of six logos is table stakes, not proof. What moves skeptics is a number attached to a name: role, company, measurable outcome.
Audit check: does your page contain at least one piece of proof with a real number in it? If every testimonial is "great product, highly recommend," you don't have proof. You have decoration.
4. Signup friction: the card question
This is where the First Page Sage table earns its keep, because the data cuts both ways. Requiring a card drops visitor→trial from 8.5% to 2.5%. That's roughly 70% of your trials gone. But it raises trial→paid from 18.2% to 48.8%, because everyone who gets through actually intended to buy.
Neither is "correct." The audit question is whether your form friction matches your model on purpose, or by accident. A card wall plus a "try it free, no commitment" headline is an accident.
And friction isn't only fields. When I launched pages.report, I kept rewriting hero copy while conversion stayed flat. Then I swapped the magic-link signup button for "Continue with Google." Login rate tripled. One button. Speed counts too: Google and Deloitte measured a 21.6% lift in lead-gen form-page progression from a 0.1s mobile speed improvement (2020, dated but directionally solid).
5. CTA matched to buyer stage
"Book a demo" and "Start free trial" are not interchangeable. A demo CTA suits high-ACV products where the buyer needs a human before committing. A trial CTA suits self-serve products where the buyer wants to touch the thing today.
The audit check: does your primary CTA match how your customers actually bought? Pull your last 20 closed deals. If 18 started with a trial and your hero says "Request a demo," you've built friction into your best path. If your buyers need security reviews and procurement, a naked trial button is skipping steps they can't skip.
When to get human eyes on it
Run the five checks yourself first. Most founders find at least one obvious leak in an afternoon, and the funnel table tells you which stage deserves the effort.
The limit of self-auditing is that you can't see your own page as a stranger. You know what the product does, so the hero always "reads fine" to you. That's the point where a $99 landing page audit makes sense: I record a video teardown of your page against these exact checks and send it within 48 hours. It's the same process, minus your blind spots.
Either way: benchmark the stage, find the leak, then change things. In that order.
– Luke
