Variantix / product walkthrough
From the first estimate to a change you publish.
A practical guide through the real Variantix flow: plan a test, choose a tier, verify payment, configure four variants, connect tracking, review evidence, and apply the winning page in your own site or CMS.
Estimate the test
Use daily eligible traffic and your baseline conversion rate.
The public sample-size calculator works without login and is free at every tier. It estimates sample size and time; it does not promise a result or make the tier choice for you.
Daily eligible visitors
Enter the traffic that can enter this experiment.
Baseline conversion
Enter the current conversion rate as a percentage.
Minimum detectable lift
Choose +5%, +10%, +20%, or +30%.
- Visitors per armEstimated sample for each variant.
- Total visitorsAll four arms combined for the current experiment setup.
- Estimated daysSample divided by eligible daily traffic.
The estimate uses two-tailed α = 0.05 and 80% power. Reliable testing needs enough observations, and traffic determines how quickly they arrive. One month may be enough to run some tests but may not provide enough evidence for a confident conclusion; up to three months can be a useful runway, not a required wait or a guarantee. Existing significance and evidence checks determine readiness. Empty or zero required inputs show dash placeholders; out-of-range entries are bounded or ignored by the calculator.
Traffic suggests a tier; compare before choosing.
- < 1,000 / daySmall Business
- 1,000–9,999 / dayModerate
- 10,000–99,999 / dayMid-Market
- 100,000–999,999 / dayGrowth
- 1,000,000+ / dayEnterprise
Compare and choose
Use the traffic suggestion as a starting point, then compare the published tiers.
The pricing page keeps the monthly/annual switch and purchase buttons tied to the same tier catalog used by checkout. Prices below are in USD and shown in the published order.
$199 / month
$1,990 prepaid / year
One experiment at a time
- AI-generated copy and layout variants
- One experiment running at a time
- Weekly stakeholder digest
- Testing runway
- Up to 3 months
- Rate lock
- Locked for 12 months / through your contract
$600 / month
$6,000 prepaid / year
Multiple concurrent experiments
- Multiple simultaneous experiments
- Autonomous traffic split + significance testing
- Weekly results digest
- Testing runway
- Up to 3 months
- Rate lock
- Locked through your first contract
$1,000 / month
$10,000 prepaid / year
Up to 5 concurrent experiments
- Autonomous traffic split + significance testing
- Up to 5 concurrent experiments
- Governed winner rollout controls
- Monthly strategy review
- Testing runway
- Up to 3 months
- Rate lock
- Locked through your first contract
$2,500 / month
$25,000 prepaid / year
Up to 10 concurrent experiments
- Up to 10 concurrent experiments
- Custom protocols for high-traffic arms
- Bi-weekly strategy review
- Testing runway
- Up to 3 months
- Rate lock
- Locked through your first contract
$4,000 / month
$40,000 prepaid / year
Unlimited experiments
- Unlimited experiments with custom protocols
- Quarterly results review with a Variantix lead
- Dedicated success contact
- Testing runway
- Up to 3 months
- Rate lock
- Locked through your first contract
Annual billing is prepaid at the equivalent of 10 monthly payments, a 16.7% saving. Small Business is rate-locked for 12 months or through the contract; the other tiers are locked through the first contract.
A 10–30% surcharge is applied manually only for top-spec requests on Moderate through Enterprise. Small Business never has this surcharge. Traffic bands are guidance, so compare capacity, features, and the full terms before choosing.
Purchase and verify
The pricing choice carries through sign-in and hosted checkout.
Select a tier and monthly or annual interval on Pricing. Variantix keeps that choice through sign-in or account creation, then opens the platform's hosted payment page when billing is available.
Choose
Pick a tier and interval on /pricing. The browser sends the tier id and interval, never a price.
Authenticate
Sign in or create an account. The return link keeps the selected tier and billing interval.
Confirm
Hosted checkout returns to a verification screen. Paid access starts only after the payment is verified for that account.
A redirect alone is not proof of payment.
This is a real external hosted payment flow, not a simulated purchase. Cancel returns to the selected tier and interval. Failed checkout offers a retry; unavailable billing says no charge was started and keeps the selection visible. If verification is delayed, the return page stays unconfirmed and can be refreshed. Pricing is computed server-side from the published tier catalog.
Set up a test
Configure four variants, review the draft, and start it from the detail page.
A signed-in account with verified paid access can create an experiment. The customer supplies the tested pages and real destinations; setup does not create customer-site content by itself.
- 1
Name and target
Name the experiment. Add 1–10 absolute http(s) target URLs, plus an optional funnel-step label and notes.
- 2
Four variants
Give each of exactly four variants a unique name, absolute destination URL, optional supplemental HTML, and allocation; all four allocations must total 100%. Choose one comparison group.
- 3
Success and rollout
Choose Signup click, Form submit, CTA click, or Purchase, then select Automatic when ready, Approval required, or Prevent rollout.
- 4
Review and start
Review the configuration and save the draft. Open its detail screen and select Start. Targeting, variants, allocation, and settings lock after launch.
Validation to expect
Non-http(s) or incomplete URLs, duplicate variant names, invalid destinations, allocations that do not total 100%, or a missing success metric are shown as field or form errors. Correct them in the wizard before creating the draft.
Connect the customer site
Install the current loader in the tested page head and send the selected goal.
After the experiment is running, open its detail screen and copy the generated script. Install it on each tested page, then call its goal hook only after the real selected action occurs.
Assignment
The loader hits GET /r/<experimentId>. Variantix records an impression for a first assignment, picks a weighted eligible variant, and keeps the assignment sticky.
Conversion
After the selected goal occurs, the customer site calls window.variantix.track(). The script posts the experiment id, goal type, and assignment token to POST /api/track.
Results appear only after these real events reach Variantix. A missing or stale head loader means no assignment impressions; a mismatched goal type is ignored, so conversions remain empty or incomplete. If the customer site enforces CSP, allow the Variantix origin in script-src and connect-src.
Read the screens and reports
Follow evidence from Overview to the experiment detail and weekly report.
The signed-in workspace uses real account data. Empty states stay empty until tracked events arrive; inconclusive results stay labeled as needing more data.
Status and setup
Draft, running, paused, or concluded; goal, selected comparison group, funnel label, and configured traffic split.
Performance
Impressions, conversions, conversion rate, lift, and confidence for each variant against the selected comparison group.
Review tools
Choose a date range, apply or reset it, refresh the results, and export the displayed results as CSV.
Evidence floor
Each side needs at least 100 visitors and 25 conversions to meet the evidence floor, and existing confidence checks still apply. Gathering enough observations takes time: one month may be enough to run some tests but may not support a confident conclusion. Up to three months is a useful planning runway when needed, not a guarantee or required wait.
The weekly report covers the last seven days of tracked events. Its active-experiment count is current, and its winners-deployed metric counts winners still deployed from a rollout in that period. AI insights may be unavailable when no summary can be produced.
Open the experiment listDecide, then apply
Choose what Variantix serves, then publish the lasting change in your own site.
Review each candidate against the selected comparison group and mark the evidence-backed winner on the experiment detail. Rollout controls change Variantix traffic routing; they do not edit customer source code.
Automatic when ready
Variantix can serve the selected winner through its experiment route once the readiness gate is met.
Approval required
Wait for the owner to accept the selected winner. Hold leaves a pending decision open; deny rejects that rollout.
Prevent rollout
Block winner deployment through the experiment route. The owner can change the policy later if rollout is allowed.
- 1. Accept, hold, or deny the proposed rollout; use rollback to return Variantix traffic to the selected comparison group.
- 2. If you keep the winning experience, copy its approved content into your own site or CMS.
- 3. Review and publish that permanent page change through your normal site workflow.
Variantix rollout affects visitors routed through /r/<experimentId>. It does not write, commit, or deploy your website source. The permanent change is a manual handoff to your team.
Ready to rehearse the journey?
Start in the public calculator, then follow the steps above against a verified account and real site integration.