AI AgentPlatform
PricingSign inBook a demo
Personyze
On your site
Website PersonalizationOne profile across every page and every session.Recommendation EngineWhat each visitor wants next, ranked by the algorithm you pick or one you write.A/B TestingA winner per audience, promoted mid-campaign rather than after it.Dynamic Landing PagesThe headline matches the ad they clicked.AI Search & Chat AgentAnswers from your own pages and catalogue, and a search box that replies in a sentence.Banners & Pop-UpsRight message, right moment, right page — without a developer.Social ProofLive scarcity, ratings and what other people just did.Meeting SchedulerVisitors book time on your own page — the meeting type, host and times picked by who they are.
Off your site
Email PersonalizationSend it from Personyze, or paste one block into the ESP you already use.Push, SMS & WhatsAppFired by what they just did rather than by a schedule.Hosted Landing PagesWe serve the whole page, on your domain or ours. No site to deploy to.
Know who they are
Behavioral TargetingSegments that update themselves from what people actually do.ABM MarketingThe company behind an anonymous visit, joined to your CRM.Website AnalyticsSessions, bounce and revenue — then click a row and there is a person.
Build & connect
MCP ServerRun your account from Claude, ChatGPT or any MCP client.IntegrationsHubSpot, Salesforce, Twilio Segment, Tealium, GTM and the REST API.APIs, SDKs & AutomationOne REST endpoint over every object, server and mobile SDKs, rules and reports that run on their own.
Pricing
A/B & Multivariate Testing

A/B testing software.
Ship the winner.

Run head-to-head tests on copy, design, layout and timing — with audience-level segmentation built in. Reach significance fast, then promote the winner automatically.

Test one element or every combination at once. Build whole landing pages and test them on your own domain. Then read the result in reporting that tells you what won, by how much, and for whom.

G24.8 Capterra4.6
Use cases

Test what matters. Ship what wins.

Hero copy, pricing pages, email subject lines, full personalization rules — run real tests on real traffic and let the data decide.

Hero & landing tests

Find the headline that converts

Test headlines, hero images, value props, and CTA copy on landing pages. Run multivariate combinations or simple A/B splits with built-in significance scoring.

  • Visual editor — no code required
  • Multi-arm tests with traffic allocation
  • Auto-promote the winner when significant
VARIANT A
2.1%
VARIANT B ✨ WIN
2.6%
+24% Avg. conversion lift on hero tests
Pricing & checkout

Optimize where revenue happens

Test pricing display, plan structure, payment flows, and trust signals. Segment by visitor type — new vs. returning, geo, source — so you don’t blow up returning revenue.

  • Audience-level test isolation
  • Revenue-per-visitor as primary metric
  • Auto-stop on significant negative lift
STARTER
$19
PRO
$49
+11% RPV
SCALE
$99
+11% Avg. revenue per visitor improvement
Email tests

Subject lines, content, send time

Test subject lines, sender names, hero blocks, and CTA copy. Open-time rendering means winning content stays fresh even after send. Auto-promote winner mid-campaign.

  • Subject-line A/B with auto-winner promotion
  • Content blocks tested per segment
  • Send-time optimization per recipient
Don’t miss this week’s deals
21%
Sara, your custom picks are ready
25%
Auto-promote winner
LIVE
+19% Avg. open rate lift on subject tests
Personalization rules

Prove the lift before scaling

A/B test personalized vs. generic experiences on the same audience. Compare two personalization strategies head-to-head. Roll out with confidence, not faith.

  • Hold-out groups baked into the campaign
  • Per-segment lift breakdowns
  • Statistical guardrails on every roll-out
New visitors
Variant A wins
+15%
Returning
Variant B wins
+22%
Mobile
Variant C wins
+9%
Hold-out
Control group
baseline
+27% Avg lift vs. generic campaigns
Test types

Pick the test. Build it without code.

A/B for one decision, multivariate when several things change at once, and a holdout that keeps part of your traffic on the untouched site so the lift you report is the lift you caused.

Split itYour variation, and the untouched site held back to measure against
Then let it decideFour strategies, or your own thresholds — the winner promotes itself
Or vary several things at onceEvery combination measured, and the effect of each group on its own
Test typeA/B testMultivariateViewFullCompactAssign by Rotate users
Traffic splitA · 50%OG · 50%
Goal & auto-pickGoal Not set● Auto-pick OffConfigure ▾
Campaign content100%

Shown to every visitor matching targeting, in any test group.

+Add actionor drop here
ARotation group A50%
● LIVE
Free delivery this weekOn every order over £40Shop now
Hero banner · free delivery
Popup, Banner & HTML Builder · v3
21,700Impressions18,750Confirmed views112Clicks · 0.52%0.73%Visitors
No own max showAdd
Site original50%

Your unchanged site — no A/B actions here. Campaign content still applies.

Test typeA/B testMultivariateViewFullCompactAssign by Rotate users
Primary goalProduct purchasesDemo signupOpen accountClicksBounce rate+ More goals…
Promote the variation that wins on your goal once it reaches 98% confidence — but hold the call until at least 1,000 sessions and 0 days have passed.

Decision strategy — how sure we should be before calling it. Higher = slower but safer.

Cautious99% · 5,000 · 21d · ≥1%
BalancedRec95% · 1,000 · 14d · ≥0.5%
Fast90% · 500 · 7d
CustomSet thresholds
Confidence 98 %Min. sessions 1,000Min. time 0 days
Readiness to decide — how close the test is to meeting all three conditions.
Confidence78%Sessions920 / 1,000Timemet
Auto-adjust rotation — gradually shift the rotation toward the likely winner while the test runs.
Give the winner all tested traffic, pause the rest — a control group, if you kept one, stays held back so the campaign can still be measured.
Test typeA/B testMultivariateViewFullCompactAssign by Random
2 groups4 combinationsEach group is one thing you’re varying. A visitor sees one action from each group — Personyze combines them into every variant below, so you can measure each group’s individual impact and detect interaction effects, not just “A vs B”.
Group 1 · headline
Headline AHeadline B
Group 2 · call to action
CTA greenCTA red
4 combinations · 2 × 2 = 4 · ~1,000 sessions each = ~4,000 sessions before a winner can be called.
#Group 1Group 2TrafficVisitorsConv. rateLift vs control
C1Headline ACTA green25%6,0412.10%control
C2Headline ACTA red25%6,1122.44%+16.2%
C3Headline BCTA green25%5,9882.31%+10.0%
C4Headline BCTA red25%6,0202.87%+36.7%
Main effect of each groupMeasured on conversion rate
Headline+12.4%
Call to action+19.8%

Which change did the work — not only which pair won.

Build variants visually

Design every variant without writing code.

Edit a live page directly — change copy, swap images, redesign CTAs — or build popups, banners, and forms from a library of templates. Every variant goes from idea to live in minutes, no devs.

Live product UI — visual editor on a real page
20+
Content widget types
15%
Avg conversion lift
4.6/5
G2CapterraCrozdesk
Average rating on G2, Capterra & Crozdesk
Methodology

Statistical rigor, without the statistician.

The thresholds are already set when you open a test, and they are stricter than the ones most tools ship with. The AI proposes the numbers for this particular campaign, you change them or leave them, and the winner promotes itself when it has earned it. No spreadsheets.

01
Step 1 · Set

The bar is already high when you arrive

Tests ship demanding 98% confidence and a floor of 1,000 sessions before anything may be called a winner, plus a minimum runtime so a good Tuesday cannot win on Tuesday. Most tools ask for 95% and let you stop whenever you like.

Presetconfidence98%· sessions1,000
02
Step 2 · Aim

You decide what winning means

Point the test at any of your twelve goal events, or at purchases per session, order value, clicks, bounce rate, pages per visit or time on site. A test that optimises for clicks and a test that optimises for revenue do not always agree — so you say which one you meant.

Winner onrevenue per session· not clicks
03
Step 3 · Judge

A win has to be worth having

Set the smallest improvement you would actually act on, per goal. A variant that is ahead by a statistically real but commercially pointless margin does not get promoted, and does not get to waste a quarter of your traffic proving it again.

Rulelift under2%→ not a winner
04
Step 4 · Promote

Auto-promotion when the winner is real

When the variant clears the confidence bar, the session floor, the runtime and your improvement rule, traffic moves to it — all of it, or the share you nominate, or none at all if you would rather keep the split and be emailed instead.

Promoteat98%· after1,000sessions → ship it

A holdout is just a split that does not add up

Variants are groups with a traffic percentage each. Give two variants 45% apiece and the remaining 10% see nothing at all — that is your control, and it keeps running for as long as the test does, so the lift you report is the lift you caused.

A · 45%B · 45% HELD OUT · 10%

And it says when two tests are fighting

Before you launch, the editor shows you which other campaigns are aiming at overlapping audiences on overlapping pages. Two pop-ups arriving together is not a result, it is a collision — and you find out while you are still editing rather than from the numbers a fortnight later.

Data in, results out

Split on any signal. Score in any tool.

One snippet installs on Shopify, WordPress, Magento or a custom stack. From there Personyze reads the audience data you already collect, so a test can be split on who someone actually is — and writes the result back to the analytics you already trust.

Audience data in
HubSpotHubSpot
SalesforceSalesforce
SegmentSegment
GA4GA4
6sense6sense
TealiumTealium
ZohoZoho
ZoomInfoZoomInfo
ClearbitClearbit
ShopifyShopify
WordPressWordPress
LeadrebelLeadrebel
+ Geo, weather, intent, lifecycle, behavior, custom attributes — or roll your own with our REST API.
Results out
Google AnalyticsGoogle Analytics 4SegmentTealiumGTM data layerKlaviyo

Every test can fire its own event to the tools you already run, per campaign or as the account default. Each tool has to be on the page already — Personyze never loads one for you, and a missing one is skipped silently.

Test results

Read the test. Defend the call.

Per-variant lift, confidence intervals, segment breakdowns, and a clear decision criterion — so you can show exactly why a variant won and roll the call past skeptics.

Live product UI — lift, confidence, segment breakdowns
FAQ

Direct answers. No fluff.

The questions that come up before signing on. Send anything else our way and we’ll answer fast.

How does A/B testing in Personyze work?

Visitors are split between variants based on the traffic allocation you set — 50/50 for a classic A/B, even slices for an A/B/n test, or 90/10 with a holdout to measure incremental lift. Each visitor gets a sticky assignment so they always see the same variant.

Personyze tracks impressions, clicks and goal events in real time, and holds a winner back until it clears the confidence threshold, the session floor, the runtime and your minimum improvement and by how much. When statistical confidence (probability that B beats A) crosses a threshold and a minimum runtime has passed, you can auto-promote the winner — or review the full breakdown by audience segment, device, and channel before deciding.

Do I need to write code to run A/B tests?

No. Personyze is built to be no-code: the visual editor lets you change copy, swap images, redesign CTAs, or replace whole sections by clicking on them on the live page. Popups, banners, and forms are built from a 20+ widget template library — no engineering team required.

For the rare cases where you want bespoke logic (custom JavaScript, server-rendered variants, integrations with proprietary data) there\'s a JS action you can drop in and a REST API for server-side variant assignment. Most teams never need either.

What platforms can I run A/B tests on?

Personyze A/B testing is fully platform-agnostic. It works on Shopify, Magento, BigCommerce, WooCommerce, Salesforce Commerce, WordPress, Webflow, custom React or Vue apps, Next.js, headless setups, and plain HTML sites. Integration is a single JavaScript snippet.

For server-side variants there’s a REST API. For email A/B tests, Personyze injects content at email-open time. And for data, Personyze plugs into your existing martech: Google Analytics 4, Mixpanel, Segment, HubSpot, Salesforce, Marketo, ActiveCampaign, and any tool with a REST API or webhooks.

What’s the difference between A/B testing and split testing?

The terms are used interchangeably — both describe splitting traffic between two or more variants and measuring which performs better. “A/B testing” is the more common phrasing for landing-page and CRO experiments; “split testing” is sometimes used to specifically describe URL-based tests where each variant lives on a different page.

Personyze handles both: same-page A/B variants via the visual editor, and full-page split tests where you redirect a percentage of traffic to a different URL. Multivariate (A/B/n) and holdout tests are also supported in the same UI — no separate tool required.

How fast can I launch a campaign?

From a fresh account, most teams launch their first campaign in under 15 minutes: pick a template, choose a trigger, define the audience, hit publish. No code, no developer queue.

For more complex setups (custom CRM data, ABM IP lookups, multi-step flows) plan on a few hours or schedule a free implementation session with our success team.

Will it slow down my site?

No. The Personyze script is under 30KB gzipped, loaded asynchronously, and decisions are made in under 50ms. We don’t block render. Customers consistently report no measurable impact on Core Web Vitals after deploying.

How does targeting by company (ABM) work?

Personyze includes built-in IP intelligence that resolves anonymous traffic to a company name when possible — with industry, employee count, revenue tier, and HQ location.

You can also push known visitors via your CRM (HubSpot, Salesforce) or a custom data layer. Combine company attributes with on-site behavior to fire campaigns only when the right account hits the right page.

Can I use my own design / brand fonts?

Yes — every popup, bar and banner is built from standard HTML/CSS, so you can use the visual editor, or drop in your own custom HTML and inherit your site’s CSS. Though you rarely need to: Personyze offers to apply your brand to any template it shows you — your colours, your fonts and your language — so a design starts on-brand instead of being restyled into it.

How do you prevent over-popping (campaign fatigue)?

Frequency limits are first-class: session caps, cooldowns, cross-campaign rules, and goal-based suppression. Set "max 1 popup per visitor per session" and Personyze enforces it across all campaigns automatically.

Once a visitor closes or converts on a campaign, that campaign is suppressed by default until you decide otherwise.

Does it integrate with my email/CRM?

Yes — native integrations for HubSpot, Salesforce, Mailchimp, Klaviyo, ActiveCampaign, and Marketo. Captures from popup forms can be auto-pushed to your CRM as new leads, with custom field mapping. Webhooks let you connect anything else.

Are popups bad for SEO?

Personyze popups are compliant with Google’s mobile interstitial guidelines by default. We never auto-show fullscreen modals on first page-load on mobile. Suppress on legal/cookie pages, gate behind scroll/dwell triggers, and keep ad-of-page-content ratio in check.

Does A/B testing work with the rest of Personyze?

Yes — testing is built into every capability rather than bolted on. Test recommendation algorithms against each other, pit popup offers and triggers head-to-head, and prove which landing page variants win for each audience.

Get started

Stop guessing. Start testing.

Launch your first A/B test in 15 minutes. Real significance, real guardrails, no statistician on staff required.

Free plan available · 30-day money-back · No annual lock-in