Unbounce With Causality Engine: Credit Past the Landing Page

The short answer: a landing page can win its A/B test at 95% confidence while the store's orders stay flat, and both screens are telling the truth. Unbounce, the landing page builder with A/B tests and AI traffic routing, counts a click or form submission on any element ticked as a conversion goal, and all goals add up to one total (conversion goals article; 95% from the results article). That conversion is a step towards the store, not an order. The budget question is which ad channel GA4 credits once the visitor has left the page and paid. Causality Engine, a causal read of the store's GA4 export, answers it per channel, starting with a free first finding.

Disclosure: I founded Causality Engine and own it, and this page recommends it. Every claim about it links to Causality Engine's own pages.

What to settle before the next budget meeting:

1. Divide the page's number by the store's. For example, 2,371 goal clicks against 214 purchases from the same visitors is 2,371 ÷ 214 = 11.1 clicks per order. Report the 214.
2. Use the switch Unbounce already ships. Pass through URL Parameters, on a button's confirmation settings, carries the ad's UTM tags on to the store link (parameters article).
3. Check one row in GA4. Purchases under Paid Social or Paid Search are the count to set beside Shopify's orders.

The page and the read, row by row

How Unbounce scores a page

On the Goals tab of the Classic Builder you tick links, buttons or forms; each completion counts in the Page Overview. The conversion goals article adds two rules: several goals all land in one total, because the reporting cannot yet separate them, and several goal clicks from one browser count once, because tracking runs on an Unbounce cookie. The Page Overview guide shows visitors, conversions and conversion rate, exportable as CSV.

A purchase can become the conversion instead. Untick every goal, then install the External Conversion Tracking script on the order confirmation page. When it loads, the script looks for the Unbounce visitor cookie and, if it finds one, adds 1 to the page's count (external tracking). For Shopify the script goes into a custom pixel under Customer events, set to fire when checkout completes (Shopify article). Neither article names a window. Any purchase from a browser that carries the cookie is credited to the page, whatever brought the buyer back.

Tests are scored on that same count. The A/B Test Center runs a chi-square test and treats 95% confidence or better as enough to promote or discard a challenger (results article). Smart Traffic reports a Lift Percentage: its conversion rate against the rate Unbounce's statistical model estimates random routing would have produced (Smart Traffic FAQ).

Keeping the ad's channel on the purchase

The page gets the click, the store gets the money, and GA4 must see one session or the ad loses the sale.

• GA4 on the page. Script Manager has a built-in Google Analytics 4 option: paste the Measurement ID, choose the domain, republish, and data arrives about 24 hours later (GA4 article). Use the store's measurement ID.
• Tags on the outgoing click. Tick Pass through URL Parameters and the button forwards whatever utm_source, utm_medium and utm_campaign the ad brought (parameters article). Unbounce writes no tags of its own (referral sources article).
• Domains. On a subdomain of the store's root, Google does not treat the page as a new referrer (unwanted referrals). On a different root domain, GA4 counts two users and two sessions unless both sites share one tag ID under Configure your domains (cross-domain measurement). Without that, the store records the page as a referral and the ad drops out.

Google files a click under Paid Social when the source is a social site and the medium contains cp, equals ppc or retargeting, or starts with paid (default channel group). Tag Meta traffic with a medium such as "landing-page" and it fails that test; with facebook as the source, the visit drops to Organic Social.

A worked example: one Meta campaign through a click-through page

For example, with figures invented for the arithmetic and no store behind them: a click-through page on promo.example.com sends Meta traffic to a Shopify store on www.example.com in October.

• The Page Overview shows 18,240 visitors and 2,371 conversions on the button goal: 2,371 ÷ 18,240 = 13.0%.
• With the goal unticked and External Conversion Tracking on the Shopify pixel, the page counts 214 purchases.
• Meta Ads Manager credits the campaign with 263 purchases by its own attribution setting.
• GA4, with Pass through URL Parameters on, gives Paid Social 151 purchases from those visitors on last click.

Meta claims 263 - 151 = 112 more than GA4 grants. Unbounce claims 214 - 151 = 63 more: buyers who carried the page's cookie but whose last click was an email, a search or a typed address. Each is right by its own rule; none says what the campaign caused.

That is the question a causal read takes. Per the how-it-works page, it splits Direct back to the channels that sent those buyers, sets Paid Social's caused share beside the last-click rebuild of the same export, grades the read with a data-health score from 0 to 100, and names a next step tested in one region first. If Paid Social's share lands well below Meta's 263, cut in one region for the days the read names, not across the account.

What the read leaves to Unbounce

Causality Engine and Unbounce have no connection. The integrations page lists no page builders and says the read covers every channel a GA4 export records. A landing page is not a channel. Which headline sold stays a question for the A/B Test Center.

It also needs material to work with. The how-it-works page asks for 12 months of GA4 data or more, and calls a channel under 1% of credit on both methods too small to act on. Visits with no tracked source come back flagged as Unassigned, with the advice to fix tagging first (pricing for agents).

The Lift Percentage that sells AI traffic routing, a feature the pricing page lists on the Optimize plan, is measured against a random-routing baseline that Unbounce's own model estimates, so the plan and the proof of its value come from the same company.

The free check: the page against no page

Unbounce's own tests compare page with page. The question here is whether the page earns its place between ad and store, and that is a by-hand split:

1. Duplicate one Meta ad set. Point one copy at the Unbounce page with Pass through URL Parameters on, the other straight at the product page.
2. Tag both with the same source and medium and a different utm_content, so GA4 keeps them apart under Paid Social.
3. Run them on equal budgets for two weeks and count purchases per utm_content in GA4, then check the total against Shopify's orders.

If the page arm sells no more than the direct arm, the Page Overview's conversion rate measures clicks, not sales.

For every channel at once, Causality Engine, which I founded, takes the GA4 Attribution paths export on the start page. The preview runs in the browser; the full read costs €99 once, refundable within 30 days (pricing, checked on 11 November 2026).

See also

• Unbounce: the builder, its plans, traffic caps and complaints
• How to use Unbounce: setting up Smart Traffic on one page
• Unbounce on the marketplace: Unbounce's affiliate program terms
• Causality Engine: what the read does, costs and declines
• How to use Causality Engine: from GA4 export to a regional test
• Causality Engine on the marketplace: the owner's own affiliate terms
• Test landing page variants with a holdout: a split with a control
• Swipe Pages vs Unbounce: AMP speed or visitor routing

FAQ

Does Causality Engine read Unbounce's conversion data?

No. Nothing passes between them. The read takes the GA4 export, where the page's visitors appear under whichever ad channel their tags name (integrations).

Why does Unbounce count more purchases than GA4 gives the ad?

External Conversion Tracking credits any purchase from a browser that holds the Unbounce cookie (external tracking). GA4's last click gives the order to whatever brought the buyer back.

My Unbounce page is on a different domain from the store. What breaks?

External Conversion Tracking, which needs a shared root domain, and GA4's session, which splits in two unless cross-domain measurement links both sites under one tag ID (Google's guide).