Migrating Off VWO Without Losing Test History
A practical playbook for switching off VWO or Optimizely while preserving every past experiment, decision rationale, and in-flight test — so consolidation doesn't cost you years of CRO learning.
Quick answer
You don't lose test history if you export it before cancelling. Pull VWO or Optimizely's report CSVs and screenshots of every concluded test, archive them as a structured learning library, bridge any in-flight test by finishing it on the old tool while the new snippet runs in shadow mode, then cut over. Budget two weeks; the audit work is bigger than the technical migration.
Migrating off VWO without losing test history
The process of switching from VWO or Optimizely to a new experimentation platform while preserving past results, decision rationale, and active tests.
Migrating off VWO or Optimizely is rarely blocked by technical complexity — the snippet swap takes an afternoon. The real blocker is institutional knowledge: two or three years of concluded A/B tests, the hypotheses behind them, and the wins and losses that shaped the current site. Cancel the subscription and that history disappears with the login.
A clean migration treats the test archive as a first-class deliverable. You export every concluded experiment, document why each decision was made, bridge any in-flight tests cleanly, and stand up the new tool with the historical context already loaded. Done well, the new platform launches with a richer learning library than VWO ever showed you.
The fear is reasonable. Most teams have 30-80 concluded tests sitting inside VWO and no copy of the results anywhere else. Once you cancel, the dashboards go dark and the reports are gone in 30-90 days depending on plan.
Why teams stall on the migration
The stack audit usually concludes that GA4 plus Hotjar plus VWO is doing the work of one unified platform at three times the cost. Consolidation makes obvious financial sense. The migration still gets postponed for two reasons.
First, no one wants to be the person who lost two years of CRO learnings. Second, there's usually one in-flight test that someone is emotionally attached to — a checkout variant, a PDP layout — and stopping it mid-flight feels like wasting weeks of traffic. Both objections are solvable in an afternoon of planning.
Data retention deadlines bite
VWO Growth plan retains report data for 90 days after cancellation; Optimizely Web's retention varies by contract. If you've already given notice, export this week — not next month. Screenshots count as a valid backup; raw CSV is better.
What to export before you cancel
For every concluded test, capture six things: hypothesis, variants (with screenshots), traffic allocation, primary and secondary metrics, statistical result, and the decision that followed. VWO's CSV export gives you the metrics half; the hypothesis and decision usually live in Notion, Jira, or a forgotten Slack thread.
Combine them into a single sheet — one row per test — with columns for tested page, hypothesis, lift, significance, and outcome (ship / kill / iterate). This becomes the learning library your new tool imports on day one. A clean library of 40 past tests is more valuable than 200 unstructured VWO reports no one can find.
Bridging in-flight tests cleanly
Don't stop a test mid-flight to swap tools. Let VWO finish the experiment on the original snippet, then ship the winner before the migration cutover. If the test is more than two weeks from significance, kill it — the variant data is worth less than the migration timeline.
While VWO finishes, install the new snippet in shadow mode: tracking only, no test activation. This validates the data layer against GA4, catches consent-mode issues, and lets your team get comfortable with the new UI before any traffic is split. Two weeks of shadow mode prevents the most common migration regret — discovering a tracking gap after the old tool is gone.
The GA4 historical import advantage
If the new platform supports GA4 historical import — pulling 12-24 months of analytics on day one — you skip the cold-start period entirely. The audit runs against real traffic from day one, segment baselines are already calibrated, and your first hypotheses come from existing drop-off patterns, not a month of waiting for fresh data.
The 14-day migration timeline
Days 1-3: export every concluded VWO test into the structured learning library. Days 4-7: install the new snippet in shadow mode and reconcile against GA4. Days 8-11: let in-flight tests finish or be killed; ship winners. Days 12-14: activate the new platform, archive the library inside it, cancel VWO.
Most Shopify teams complete this without a developer — the snippet installs through the theme editor and the GA4 import is a one-click OAuth. The audit work (curating the learning library) is the constraint, not the engineering. Block half a day from whoever owns CRO and protect it.
Frequently asked questions
VWO retains report data for approximately 90 days after subscription end on most plans, though enterprise contracts vary. Export to CSV before you give notice — don't rely on the grace period. Screenshots of each test's results page are a valid secondary backup if CSV export is restricted on your tier.
No platform offers a one-click VWO import because the schemas differ and VWO doesn't publish a portable export format. The practical path is a manual structured library — one row per test, indexed by page and outcome. This is more useful than a raw import anyway, because most VWO archives contain dozens of inconclusive tests that clutter the new tool.
Let them finish on VWO if they're within two weeks of significance. Run the new platform's snippet in shadow mode (tracking only) alongside VWO during this period so you validate data parity before cutting over. Tests more than two weeks out should be killed; the migration timeline is worth more than marginal variant data.
No, provided the new snippet runs in tracking-only mode (no test activation) during the shadow period. Two tracking snippets on the same page is standard during any tool migration and won't cause flicker or layout issues. Only activate experiments on the new tool after VWO is removed from the theme.
A platform that supports GA4 historical import gives you 12-24 months of analytics on day one, so the new tool launches with calibrated baselines rather than a cold start. See GA4 historical import: day-one audit vs cold-start tools for the full comparison. This usually shortens time-to-first-test from six weeks to under one.
The playbook is identical: export concluded tests, bridge in-flight tests in shadow mode, then cut over. Optimizely's export tools are slightly more generous on enterprise contracts (raw event data is sometimes available), but the structured learning library is still the deliverable that matters. Don't get distracted by raw data dumps you'll never query.
On Shopify or WooCommerce, usually not — the snippet swap happens in the theme editor and modern tools auto-detect the platform. You'll want developer time only if you have custom tracking events, a headless front-end, or server-side experiments. For the typical Shopify store running standard VWO Web, the migration is a marketing-ops task, not an engineering project.
Most teams in the €1M-€15M revenue band pay €600-€2,400 per month combined across VWO, Hotjar, and any GA4 add-ons. A unified platform typically lands at 40-60% of that, with the bigger gain being team time saved on reconciling three dashboards. See GA4 plus Hotjar plus VWO vs unified CRO platform for the breakdown.
Cancelling before the export is done. The second-biggest is exporting raw CSVs without the hypothesis and decision context — the numbers alone don't tell you why the team killed the free-shipping test. Capture the rationale while the institutional memory is still warm; six months later no one remembers.
Avoid it. The shadow-mode bridge is safe, but cutting over the active testing tool two weeks before peak introduces risk you don't need. The ideal migration window is January-February or July-August — low-stakes traffic periods where a tracking misconfiguration is recoverable. Plan the project around the calendar, not the contract renewal date.
See Metricuno on your data
Bring your stack — Google Analytics, Stripe, a CRM, anything — and we'll walk through the metric tree that turns your funnel into one number.