Here's a conversation that happens more than it should.
A SaaS company hires a "CRO agency." Three months later, they have a redesigned homepage, a new color scheme for the CTA buttons, and a final report full of screenshots from Hotjar. CVR has moved from 2.1% to 2.3%. The agency calls it a win. The founder isn't sure what they paid for.
This is the gap between what most agencies sell as CRO and what conversion rate optimization actually is. The difference matters - not just philosophically, but in terms of what you spend and what you get back.
This guide is for SaaS founders, VPs of Marketing, and growth leads who know conversion is a problem and are trying to figure out what kind of help they actually need. We'll cover what a genuine CRO agency does, how to evaluate one, what questions to ask before signing anything, and what red flags to walk away from.
What CRO Actually Is (And Isn't)
Conversion rate optimization is not a synonym for A/B testing. It's not landing page redesign, button color testing, or "making things pop." Those are tactics - and tactics without a strategic foundation are just guessing with extra steps.
Real CRO is a structured process - and according to Econsultancy, only 28% of companies are satisfied with their conversion rates, yet the majority still lack a structured testing programme. Real conversion optimization improves the website's conversion rate and can increase revenue without increasing traffic. The gap is almost always between tactics and process.
Discovery and diagnosis. Before touching a single element, a real CRO agency starts with data. Quantitative analysis (Google Analytics, session recordings, funnel drop-off data) combined with qualitative research (user interviews, heatmaps, on-site polls) to understand why visitors aren't converting. The question being answered is: where is the friction, and what's causing it?
Hypothesis formation. Every test should start with a specific, falsifiable hypothesis: "We believe that moving social proof above the fold will increase CTA clicks by X% because visitors are expressing doubt about credibility before they've seen the product's customer base." Agencies that skip this step are running experiments, not CRO.
Prioritisation. Not every hypothesis gets tested. Good CRO agencies prioritise tests by potential impact, confidence in the hypothesis, and ease of implementation - commonly called a PIE or ICE framework. This is how you get the highest-return experiments running first.
Statistical rigour. A test result is only meaningful if it's statistically significant. An agency that declares a winner after three days and 200 visitors doesn't understand testing. Real CRO requires defined confidence thresholds (typically 95%), adequate sample sizes, and an understanding of how to avoid false positives.
Iteration. CRO is not a project - it's a programme. Each test, win or loss, generates insight that feeds the next hypothesis. An agency that pitches you a "three-month CRO project" and walks away at the end hasn't understood this.
Tools like VWO, Optimizely, and Microsoft Clarity provide the infrastructure for testing and research - but the thinking behind the experiments is what separates a real CRO engagement from expensive guessing.
What a Real CRO Agency Actually Does
A genuine conversion rate optimisation agency operates as a research and experimentation function, delivering CRO services and conversion rate optimization services through a multidisciplinary team that typically includes UX/UI designers and data analysts, not a design function. Here's what that looks like in practice.
Conversion audit. The starting point for any engagement is a thorough audit: analytics data, session recordings, heatmaps, exit surveys, and - critically - existing customer interviews. The audit surfaces where the biggest leakage points are and gives the team something concrete to hypothesise against.
Conversion architecture review. Before running a single test, experienced CRO practitioners review the page structure itself: does the hero answer "what is this and who is it for" within five seconds? Is social proof placed where doubt is highest? Is the CTA friction-matched to where the buyer is in their decision? Many of the highest-impact changes come from structural problems that don't require A/B testing to identify. For a deeper breakdown of what this looks like on SaaS landing pages, see our guide to the best SaaS landing pages.
Hypothesis backlog. The agency builds and maintains a prioritised backlog of test hypotheses, each tied to specific data points. The backlog is a living document - new data feeds new hypotheses, and wins or losses from live tests move hypotheses up or down the priority list.
A/B and multivariate testing. Tests run against defined traffic segments, with adequate sample sizes and confidence thresholds agreed in advance. Results are reported with context - not just "the variant won" but "why we think it won and what that tells us about our visitors."
Post-test analysis. A test that "loses" is still valuable. A good CRO agency extracts the learning: was the hypothesis wrong, was the sample too small, or did the test expose a different problem than the one it was designed to solve?
Iteration and compounding. Each winning test is implemented and becomes the new baseline. Each losing test informs the next hypothesis. Over a 6-12 month engagement, this compounding effect is where the real gains come from.
What Separates a Real CRO Agency From a Design Agency "Doing CRO"
This distinction is worth dwelling on because the market is full of web design agencies and digital marketing agencies that list "CRO" as a service - and the overlap is real enough that it's easy to hire the wrong thing.
Here's how to tell the difference.
A design agency doing CRO starts with the page and asks "what would make this look better?" Their output is usually a redesign - new layout, new visuals, possibly new copy. They may run an A/B test comparing old vs. new. They're good at making things look better. They are not good at identifying why visitors don't convert or running a rigorous programme of experiments.
A real CRO agency starts with the visitor and asks "why aren't they converting?" Their output is hypotheses backed by data, experiments designed to test those hypotheses, and a programme of iteration. This depends on rigorous data analysis of user behavior. They may end up recommending design changes - but the design is the output of the research, not the starting point.
The clearest indicator: ask any agency to walk you through their last five tests. A real CRO agency can tell you the hypothesis behind each test, the data that generated it, the result, the statistical confidence, and the learning - whether it won or lost. A design agency doing CRO will show you before-and-after screenshots.
At Flowout, our CRO service is built on the research-first model. We work primarily with SaaS companies where conversion architecture - not visual polish - is the lever. The work starts with a data driven approach tied to your business goals, not our design preferences.
How to Evaluate a Conversion Rate Optimization Agency
If you're in the market for a CRO agency, the evaluation process matters as much as the shortlist. Here's what to look for.
Questions to Ask
"Walk me through a test that failed. What did you learn?"
This is the most revealing question you can ask. Agencies that only show you winning tests either cherry-pick their case studies or don't run enough experiments to have a meaningful failure rate. A real CRO programme will have tests that don't move the needle - and the team should be able to tell you exactly what they learned from them.
"How do you define statistical significance, and what's your minimum sample size policy?"
If the answer is vague, walk away. Every reputable CRO practitioner has a defined confidence threshold (usually 95%) and a methodology for calculating required sample sizes before a test starts. Agencies that "just know when a test is done" are reporting noise as a signal.
"What tools do you use for qualitative research?"
Good CRO agencies will mention session recording tools (Microsoft Clarity, Hotjar, Crazy Egg), on-site polling tools, and - ideally - user interviews or usability testing. If the answer is only quantitative (just Google Analytics, just A/B testing tools), they're missing half the picture.
"How do you use advanced tools to track user behavior and identify drop off points across the conversion funnel?"
"What does your hypothesis prioritisation process look like?"
Look for a structured framework (PIE, ICE, or similar) rather than gut feel. The best CRO agencies are disciplined about where they spend testing bandwidth - not every hypothesis deserves a test, and not all winning tests are equally valuable.
"What does a typical engagement look like in months 1, 3, and 6?"
The answer tells you whether they see CRO as a programme or a project. If month 6 sounds like month 1 with more tests, that's a good sign. If they're describing a "final delivery" at month 3, you're looking at a project agency, not a CRO partner.
"Who owns project management, and how do you handle consistent communication during the engagement?"
Pricing Models
CRO agencies typically price in one of three ways:
Retainer. A monthly fee covering a defined scope - typically a set number of tests per month, research, and reporting. Retainers are the most common model for ongoing optimization and a structured experimentation program. Ranges from $3,000–$20,000+/month depending on agency size and test volume.
Performance-based. The agency takes a share of revenue uplift. Sounds appealing but creates misaligned incentives - agencies optimize for short-term metric lifts rather than long-term programme quality, and attribution gets complicated fast. Invesp and Conversion.com are among the few agencies that operate at this level with genuine rigour.
Project-based. A fixed fee for a defined deliverable - usually a conversion audit plus a set of test recommendations. Useful as a starting point but not a substitute for an ongoing programme, and often a first step before broader CRO programs.
For most SaaS companies at growth stage, a structured retainer with a clear testing cadence is the right model. See our pricing page for how Flowout structures CRO engagements.
Red Flags to Walk Away From
"We guarantee a 30% lift in 90 days."
No legitimate CRO agency guarantees specific lifts on specific timescales. Results depend on your traffic volume, your starting CVR, your existing UX quality, and a dozen other variables. Guarantees like this exist to win deals, not to set accurate expectations. That said, strong CRO efforts can still produce measurable results, and CRO services can sometimes improve conversion rates by 20% or more.
They can't show you a test that failed.
As above. Cherry-picked case studies are a signal that either the testing programme isn't real or the reporting isn't honest.
Their "CRO" is actually landing page design.
A redesigned landing page is not CRO. It might be part of a CRO programme - but if the agency's pitch is primarily visual (wireframes, mockups, redesign proposals), you're buying design, not experimentation. These are different things, and both have value - but they shouldn't be confused. For Webflow landing page design done properly alongside conversion thinking, both disciplines need to be represented in the team.
No mention of qualitative research.
Heatmaps, session recordings, and user interviews are essential inputs to good CRO hypotheses. Agencies that work only from analytics data are missing the "why" behind the numbers - which is exactly where the most valuable hypotheses come from. The Nielsen Norman Group has documented for decades how qualitative methods surface issues that quantitative data alone cannot.
They want to start with the homepage.
The homepage is rarely where the biggest conversion problems are. The highest-leverage pages for most SaaS companies are pricing pages, trial/signup flows, and key landing pages. An agency that defaults to the homepage without using data analysis to audit where traffic drops off hasn't done their homework.
What Results Look Like From a Real CRO Engagement
Two examples from Flowout's own work.
ShipHero came to us with a conversion problem: high-intent warehouse operations teams were landing on their page and not converting. The issue wasn't the design - it was the message. The hero was product-led when the audience was pain-led. After a conversion architecture review and hero restructure built around a specific, quantified pain point ("30% of the time in your warehouse is wasted"), CVR moved from 6.33% to 23%. That's not a small A/B test win - that's a structural shift in how the page communicates. Read the full ShipHero case study.
Jasper had a different problem: a crowded AI marketing tools market where vague positioning ("AI for your marketing team") wasn't differentiating them from dozens of near-identical competitors. The work focused on conversion architecture and hero messaging built around transformation rather than features - making "marketing" the specific anchor for an AI-agents-first pitch. Demo bookings increased by 62%. Read the full Jasper case study.
In both cases, the work wasn't "run some A/B tests." It was to understand the customer journey, diagnose the gap between what the page communicates and what the buyer needs to hear, and build the structural change around that insight. That's what a real CRO engagement looks like. The discipline is built on continuous testing at scale, with leading CRO agencies citing more than 52,000 A/B tests run since 2006.
How to Know If You Need a CRO Agency
Not every conversion problem requires an agency. Here's a rough decision framework, especially for SaaS and e commerce companies focused on lead generation or online sales.
You need CRO if:
- Your page has meaningful traffic (1,000+ monthly visitors on key pages) but CVR is below category benchmarks
- You've redesigned or relaunched in the last 12 months and conversion didn't move materially
- You're spending on paid advertising, including google ads, but your cost per acquisition keeps climbing
- You have high scroll depth and time-on-page but low CTA clicks - people are interested, but something is blocking them from converting
You might not need a CRO agency if:
- Your traffic is too low to run statistically valid A/B tests (under ~500 monthly visitors per key page)
- You don't have the internal resources to implement test results quickly - slow implementation kills the value of a testing programme
- Your product or positioning is still changing significantly - CRO on a product that's pivoting every quarter isn't going to compound
If you're building a custom SaaS website from scratch, it's also worth thinking about conversion principles from the ground up rather than retrofitting CRO after launch.
Working With a CRO Agency: What to Expect
A good first engagement with a CRO agency typically runs in three phases.
Phase 1 (weeks 1–4): Audit and diagnosis. The agency reviews your analytics, runs heatmaps and session recordings on key pages, conducts user research or customer interviews, and produces actionable data on conversion opportunities with the evidence behind each one.
Phase 2 (months 2–4): First tests. The highest-priority hypotheses go into the testing queue. Each test is designed, launched, monitored, and reported with context. Expect a mix of wins, losses, and "inconclusive" results - all of which inform the next phase.
Phase 3 (months 5+): Iteration and compounding. Winning tests are implemented and become the new baseline. The hypothesis backlog is updated with learnings from previous tests. This is where the compounding starts - each month builds on the last, supporting continuous improvement through funnel optimization and broader CRO strategies.
If you're working with a Webflow-native team (which matters significantly for implementation speed), our services include conversion architecture review as part of the design and development workflow - not as a separate bolt-on, and the CRO process should align with your wider digital marketing efforts. For teams using dedicated landing page tools like Unbounce, the same principles apply - the platform doesn't change what good CRO looks like.
If you want to understand whether your current pages have a conversion architecture problem, we're happy to take a look.
FAQ: Conversion Rate Optimization Agencies
What does a conversion rate optimization agency do?
A CRO agency diagnoses why website visitors aren't converting on your website and runs structured experiments to fix it. This includes qualitative research (heatmaps, session recordings, user interviews), data analysis, understanding user behavior, hypothesis formation, A/B and multivariate testing, and iterative improvement to support conversion optimization efforts. Real CRO agencies are research and experimentation practices - they're distinct from web design agencies, even though both may work on landing pages.
How much does a CRO agency cost?
Retainer engagements typically run $3,000–$20,000+/month depending on scope and agency size. Project-based audits range from $5,000–$25,000. Performance-based arrangements exist but are uncommon and come with significant caveats around attribution and incentive alignment. The right question is not "how much does it cost" but "what testing velocity and what depth of research does this budget buy?" Pricing also reflects the technical expertise behind the engagement, not just test volume.
How long does CRO take to show results?
Meaningful results typically emerge over 3–6 months - enough time to complete several rounds of testing, learn from the results, and start compounding wins. Agencies that promise dramatic lifts in 30–60 days are either working on very high-traffic sites (where test cycles are short) or overpromising.
What's the difference between a CRO agency and a web design agency?
A web design agency asks "how should this look?" A conversion optimization agency asks "why aren't visitors converting?" and bases its recommendations on structured research rather than aesthetics. The design agency's output is a redesign. The CRO agency's output is a tested, data-backed improvement to a specific metric. In practice, the best CRO work usually involves both design and research - but the research has to come first.
What conversion rate should I expect after working with a CRO agency?
It depends on your starting point, your traffic volume, and the nature of your conversion problem, but stronger conversion rates can also lead to increased revenue and a higher average order value. Industry benchmarks put average SaaS landing page CVR at 2–5%. High-performing pages can reach 8–12%+. The ShipHero result in this post (6.33%→23%) is exceptional - but the structural lever that drove it (pain-led hero for a high-awareness audience) is repeatable. That kind of CRO work can support revenue growth without necessarily increasing traffic.
How do I know if a CRO agency is legitimate?
Ask to see five tests they've run - wins and losses, because you're looking for a reliable partner, not just a vendor. When evaluating a cro company, look for a proven track record. Ask about their statistical significance methodology. Ask what qualitative research tools they use. Ask what a losing test tells them. Agencies that can answer these questions clearly and specifically, and where the right cro agency connects recommendations to business needs, are operating a real CRO programme. Agencies that pivot to design portfolios and before/after screenshots are selling something adjacent.
Does CRO work for SaaS companies specifically?
Yes - and conversion rate optimization CRO is especially effective for SaaS companies. SaaS companies typically have defined acquisition funnels, measurable conversion events (trial signups, demo bookings, paid conversions), and high enough traffic to run tests at meaningful velocity, and CRO often works alongside SEO, search engines, and social media marketing as part of a broader growth strategy. The combination of recurring revenue and compounding growth also means that CVR gains don't just affect this month's MRR - they affect every month's MRR after that, improving both conversion volume and revenue performance.
