Our pick · Verified 18 August 2026
Unbounce — landing pages and A/B testing without a developer
From $29/mo ($22 billed annually)
Most research teams do not have a CRO problem so much as a publishing problem: they cannot get a page changed without a ticket to a central web team. Unbounce solves both at once — standalone pages that live outside your institutional CMS, plus unlimited A/B testing from the Experiment tier at $149/month ($112 billed annually). For a recruitment campaign that needs eight variants live this week, that combination is worth more than any analytics feature.
Try Unbounce Opens on the vendor’s site · CASRAI referral link
Compare landing page builders in detail → — If A/B testing is not the requirement, the cheaper Build tier at $99/month ($74 annually) covers unlimited pages without it.
Editorial disclosure: CASRAI has commercial referral arrangements with some of the vendors named on this page, and may earn a commission if you subscribe to them. We name them here regardless of whether a link is present. We only recommend tools our editorial team has independently researched. Read our full disclosure policy.
In summary
- CRO tooling splits into four jobs: building pages, testing variants, watching behaviour, and capturing intent. Most teams only need the first two.
- Unbounce: Starter $29/mo, Build $99/mo, Experiment $149/mo (unlimited A/B testing), Optimize $249/mo (AI traffic routing). Annual billing is 25% cheaper. Verified 18 August 2026.
- A/B testing on low-traffic pages is usually theatre. Below a few hundred conversions per variant you cannot detect anything but enormous effects.
- Peeking at a running test and stopping when it looks significant inflates your false positive rate badly — fix the sample size in advance.
- For study recruitment, ethics approval covers your recruitment materials. Testing new ad or page copy usually means an amendment.
Unbounce plans
Verified from the vendor pricing page, 18 August 2026. Annual prices in brackets.
| Dimension | Starter | Build | Experiment | Optimize |
|---|---|---|---|---|
| Price / month | $29 ($22) | $99 ($74) | $149 ($112) | $249 ($187) |
| Pages | 5 | Unlimited | Unlimited | Unlimited |
| Traffic volume | Up to 500 | Up to 20,000 | Up to 30,000 | Up to 50,000 |
| Users | 1 | 1 | 3 | 5 |
| Root domains | 1 | 1 | 2 | 3 |
| A/B testing | No | No | Unlimited | Yes |
| AI traffic routing | No | No | No | Smart Traffic |
| AI copywriting | No | Yes | Yes | Yes |
Concierge and Agency tiers are custom-priced, starting at 100,000 and 50,000 traffic volume respectively.
The four things “CRO tool” can mean
The category is vague enough that vendors in quite different businesses all claim it. Sorting them by job makes the buying decision much easier.
1. Page builders. Create and publish standalone pages without a developer or a CMS ticket. This is the bottleneck for most research teams, and it is where the money should go first. Unbounce, Instapage and Leadpages sit here.
2. Testing and experimentation. Serve variants, split traffic, and report which performed better. Sometimes bundled with the builder (Unbounce from the Experiment tier), sometimes a separate layer (Optimizely, VWO) that overlays your existing site.
3. Behavioural analytics. Heatmaps, scroll maps and session recordings — where people look, how far they scroll, where they abandon. Hotjar and Microsoft Clarity are the common names; Clarity is free and genuinely good, which makes it hard to justify paying for this category early.
4. Capture and personalisation. Pop-ups, exit-intent overlays, sticky bars, form optimisation. Effective and easily overdone; an exit-intent overlay on a patient-information page reads badly and can undermine the trust the page exists to build.
Most research teams need categories 1 and 2, should add free tooling from 3, and should be cautious with 4.
Where this actually applies in research
Clinical trial and study recruitment. The most direct application, and the one with real money attached — recruitment shortfalls are among the most common reasons trials fail, and per-participant recruitment costs make a percentage-point improvement in conversion materially valuable. A dedicated page per study, per site, or per audience segment, with the eligibility criteria and the time commitment stated plainly, consistently outperforms a link into a departmental site.
Cohort retention and follow-up. Longitudinal studies lose participants at every wave. A well-built follow-up page — clear about what is being asked, how long it takes, and what has come of the study so far — is a cheap intervention against attrition.
Event and training registration. Conferences, summer schools, methods workshops. Institutional event pages are often buried three levels deep in a CMS; a standalone page you can publish and iterate is faster and converts better.
Survey and panel recruitment. Where the sampling frame is built rather than given, response rate is the study’s binding constraint.
Core facility and instrument bookings. A facility that wants external users needs a page that explains capability and access without requiring a phone call first.
A/B testing honestly
Researchers ought to be better at this than marketers, and are often worse — because the tool presents a confident-looking result and nobody applies the standards they would apply to their own field’s data.
Fix the sample size before you start. Decide the minimum effect worth detecting, then work out how many conversions per variant you need to detect it. The uncomfortable answer is usually large: detecting a change from a 10% to a 12% conversion rate at conventional power needs on the order of several thousand visitors per variant. A recruitment page seeing 200 visits a month cannot run this test, and pretending otherwise wastes months.
Do not peek and stop. Watching a running test and stopping the moment it crosses significance is the single most common error, and it badly inflates the false positive rate — repeatedly testing accumulating data is a multiple comparisons problem however the dashboard presents it. Either commit to a fixed sample size in advance, or use a method actually designed for continuous monitoring (sequential testing, or a Bayesian approach with a pre-agreed decision rule). Most built-in testing tools assume the fixed-horizon design; treat their significance indicator accordingly.
Watch for novelty and seasonality. A change can perform well initially because it is different rather than better. Run through at least one full weekly cycle, and be sceptical of tests spanning holidays, term boundaries or a conference.
Test one substantive thing. Changing headline, image, form length and button colour simultaneously tells you the combination won, not why. That is fine if you only want the better page, and useless if you want a transferable finding.
Measure the outcome you care about. Optimising form submissions can increase unqualified enquiries and reduce enrolled participants. If your real endpoint is consented participants, instrument through to that endpoint before declaring a winner.
When traffic is too low, stop testing and start reasoning. Below a few hundred conversions per variant, spend the effort on qualitative work instead — five usability sessions with people resembling your target participants will find more real problems than an underpowered test ever will.
Recruitment materials are usually ethics-approved
This is the point where research CRO diverges sharply from marketing CRO, and it is easy to trip over.
For a clinical trial or most human-subjects research, your recruitment materials — adverts, flyers, web pages, the wording of a screening form — are part of what the IRB or ethics committee approved. Substantively changing that copy generally requires an amendment before the new version goes live. A/B testing headlines means testing approved material against unapproved material, which is exactly the thing the approval exists to prevent.
The workable approach is to seek approval for a set of variants up front. Committees are generally receptive when the purpose is explained, because better-targeted recruitment materials serve participants too. What causes problems is testing first and asking afterwards.
Two related points. Cookie consent and analytics tracking on a recruitment page fall under GDPR, ePrivacy or their local equivalents, and a page collecting health-related expressions of interest deserves more care than a marketing landing page — decide where that data lands before you publish, not after. And be careful with urgency and scarcity tactics that are routine in commercial CRO; “only 3 places left” on a trial recruitment page is at best misleading and at worst undermines informed consent.
The changes that reliably help
Before running any test, these are worth doing simply because they are better:
- State eligibility criteria plainly and early. People self-screen. Making it easy reduces wasted enquiries on both sides.
- Say what the time commitment is. Number of visits, duration, travel required. Vagueness here is the most common reason people abandon.
- Say what happens next after they submit — who contacts them, how soon, by what channel.
- Cut the form to what you need for the next step. Screening does not require a full history; ask for contact details and eligibility essentials only.
- Make it work on a phone. Most participants will arrive on one, and institutional pages routinely fail here.
- Name the institution and the PI, with a real contact. Trust is the binding constraint on health research recruitment, not persuasion.
- Load fast. Page abandonment rises sharply with load time, and heavy institutional templates are usually the culprit — a standalone page avoids inheriting them.
Start with pages you can publish yourself
For most research teams the binding constraint is not testing sophistication — it is being unable to publish a page without a central web team. Fix that first; the testing is only useful once you can iterate.
From $29/mo ($22 billed annually)
Try Unbounce Opens on the vendor’s site · CASRAI referral link
Frequently asked questions
What are conversion rate optimisation tools?
Software for improving the share of visitors who take a desired action. They split into four jobs: building landing pages, running A/B tests, analysing behaviour through heatmaps and session recordings, and capturing intent through forms and overlays. Most teams need the first two and can use free tools for the third.
How much do CRO tools cost?
Unbounce runs from $29/month (Starter) to $249/month (Optimize), with 25% off for annual billing — A/B testing starts on the Experiment tier at $149/month, or $112 billed annually. Verified 18 August 2026. Behavioural analytics can be free: Microsoft Clarity offers heatmaps and session recordings at no cost.
How much traffic do I need to A/B test?
More than most research pages get. Detecting a change from a 10% to a 12% conversion rate at conventional power needs roughly several thousand visitors per variant. Below a few hundred conversions per variant you can only detect enormous effects, and running the test anyway mostly produces false confidence. Do qualitative usability work instead.
Can I A/B test clinical trial recruitment adverts?
Not freely. Recruitment materials are generally part of what your IRB or ethics committee approved, so testing an unapproved variant against an approved one needs an amendment first. The workable route is to submit a set of variants for approval up front — committees are usually receptive when the rationale is explained.
Why is stopping a test early a problem?
Because repeatedly checking accumulating data and stopping when it crosses significance is a multiple comparisons problem, and it substantially inflates your false positive rate. Either fix the sample size in advance and wait, or use a method designed for continuous monitoring such as sequential testing or a Bayesian rule agreed beforehand.
Do I need a landing page builder if we already have a website?
Usually yes, and the reason is organisational rather than technical. Institutional CMS platforms typically require a central web team ticket for changes, which makes iteration impossible on a campaign timescale. A standalone builder lets the team that owns the recruitment target also own the page.
