⚡
A snippet that never breaks your page
A tiny, asynchronous script with anti-flicker. If anything fails, it shows the original page and stays silent. Your conversions never depend on our infra.
A/B testing at the frontier of technology
The A/B test that respects your page, your data, and your budget.
Donnu A/B runs real experiments on your site: a snippet that never breaks the page, Bayesian statistics that call the winner honestly, and privacy by design. All at an absurdly fair price.
From US$ 29/month. No free plan: you start with full access to the product.
New vs. current checkout
9 days of data · 24,180 visitors
🏆 Variation B wins
Probability of B beating the control
99.7%
decision threshold 95%
A · control
4.80%
B
7.23% +50.8%
- < 3 KBsnippet that never blocks the page
- Bayesianan honest winner, no p-value theater
- Privacy-firstyour own cookie, zero fingerprinting
- Auditedper-account isolation + pentest
Why Donnu exists
A/B testing is usually too amateur or too expensive.
On one side, guesswork and spreadsheets: you change the page and hope. On the other, the giant suites, powerful, heavy, and charging hundreds of dollars a month for a pile of things you never use. Donnu A/B hits the sweet spot: the core of A/B testing done with rigor, fast, and at a price that fits.
The product
The core of A/B testing, done right.
No fluff. Every piece exists so you can decide with confidence without hurting the people who visit your site.
📊
Statistics that don't lie
A Bayesian engine (Beta-Binomial) that answers the only question that matters: the probability of a variation beating the control. With SRM safeguards and minimums, without encouraging peeking.
🎯
Advanced triggers
Choose exactly when and for whom the test runs: device, UTM, visitor type, tags, scheduling, and triggers by time, scroll, exit intent, click, or inactivity.
🔀
Real Split URL
Compare entire pages at different URLs, with a redirect that preserves your parameters (UTMs and the like). Traffic split and allocation under your control.
🎛️
Rich conversion goals
Nine goal types: page visit, click, form, scroll, time, video, download, code event, and API. One primary goal decides the winner, plus two secondary ones you track.
🧭
Product-led dashboard
Site registration with automatic verification, a step-by-step wizard, reports that explain the verdict in plain language. No manual, no support needed.
How it works
From zero to a winner in three steps.
1
Paste a snippet
One line on your site, once. We have instructions for WordPress, Shopify, GTM, Wix, and code. We verify the install for you automatically.
2
Create the experiment
Define the variations, the traffic split, the triggers, and the goals in a guided wizard. Publish with one click, with no redeploy of your site.
3
Let statistics decide
Follow along in near real time. When there is real evidence, Donnu calls the winner with a probability and a confidence interval. Then you just apply it.
Security and privacy
Security is not an extra. It is the foundation.
We built Donnu on the principle that your data and your visitors' data are untouchable.
🛡️
Privacy by design
A first-party cookie on your own domain, no fingerprinting and no cross-site tracking. No personal data left in URLs. Privacy is a selling point, not a patch.
🔒
Full isolation between accounts
Multi-tenant with row-level security from day one. Each account sees only its own data, validated by a security audit and penetration testing.
🧯
Fail-safe by default
If our infra is slow, the visitor sees the original page and is left out of the experiment. We never blank out your page because of us.
⚖️
Numbers you can trust
Unique counting on the server, dedupe, sample ratio mismatch (SRM) alerts. A winner is only declared when the evidence holds.
Pricing
A price so fair it makes you smile.
Everyone starts paying, with real access to the product. No free plan, no trial that turns into a surprise charge. Cancel anytime.
Standard
To validate your first hypotheses with method.
US$29/month
Start on Standard
- 2 experiments running at once
- A/B variations (control + 1)
- 2 metrics per experiment
- 1 connected domain
- You + 1 teammate
- Bayesian report with a verdict
Pro
To scale your tests with fine-grained targeting.
US$49/month
Start on Pro
- Everything in Standard, plus:
- 4 experiments running at once
- Variation C (A/B/C)
- 3 metrics per experiment
- 3 connected domains
- Up to 6 teammates
- Advanced triggers unlocked
Enterprise
Tailored for larger operations.
Custom
Talk to us
- Unlimited experiments, domains, and team
- Everything in Pro
- Dedicated onboarding and support
- Custom limits and billing
Full comparison
| Feature | Standard | Pro | Enterprise |
|---|---|---|---|
| Monthly price | US$ 29 | US$ 49 | Custom |
| Experiments running at once | 2 | 4 | Unlimited |
| Variations per experiment | A/B (2) | A/B/C (3) | Unlimited |
| Metrics per experiment | 2 | 3 | Unlimited |
| Connected domains | 1 | 3 | Unlimited |
| Team members | 2 | 6 | Unlimited |
| Advanced triggers (device, UTM, visitor, tags, schedule, triggers) | ✕ | ✓ | ✓ |
| Split URL + 9 goal types | ✓ | ✓ | ✓ |
| Bayesian report (probability, CI, SRM) | ✓ | ✓ | ✓ |
| Privacy-first security and per-account isolation | ✓ | ✓ | ✓ |
| Support | Self-service | Self-service | Dedicated onboarding |
Equivalent tools charge hundreds of dollars a month. Donnu delivers the core with the same rigor for a fraction of that.
Frequently asked questions
What people usually ask.
Is there a free plan?
No. Everyone starts paying, with full access to the product from day one. No trial that expires and no surprise charge. From US$ 29 a month.
Do I need to know how to code?
No. You paste a snippet once, with ready-made instructions for WordPress, Shopify, Google Tag Manager, Wix, or code. Everything else happens in the dashboard, step by step.
Does it work on any site?
Yes, on any site where you can add a script to the HTML. It covers the root domain and all the subdomains you register.
Is my data and my visitors' data safe?
Yes. A first-party cookie on your domain, no fingerprinting and no cross-site tracking. Accounts are isolated with row-level security, with a security audit and penetration testing. Privacy is part of the project, not an add-on.
How do you decide the winner?
With Bayesian statistics: we compute the probability of each variation beating the control and only declare a winner when the evidence is sufficient (a minimum of visitors, days, and a distribution check). No p-value theater and no encouraging early decisions.
Can I cancel anytime?
Yes. It is monthly with no lock-in. You stay in control of your account and your experiments.
Start discovering what converts today.
From US$ 29/month, with full access to the product and real security.