If reported purchases look unreliable, verify one order from checkout to the ad account before judging campaign performance. Automated bidding needs a usable conversion signal. A dashboard can show a number while the tag fires twice, omits value, or counts a checkout step as a completed sale. The store’s order record is the starting point for this audit.
Choose one business outcome
For an ecommerce campaign, write down what counts as a purchase: a confirmed order with a unique order ID, a final value, currency, and purchased items. Keep abandoned checkouts and add-to-cart events as separate actions. If the business mainly takes leads, define a verified enquiry and qualify it separately from form submissions. Our offline conversion planning guide shows how to connect later CRM outcomes to campaign review. For a B2B SaaS example, see the Google Ads demo-booking guide, then verify that the reported demo is a qualified opportunity. Avoid optimizing a purchase campaign to a convenient click or page view because those are easier to record.
Google’s GA4 purchase guidance describes the purchase event and ecommerce parameters. The transaction ID guidance explains why a unique, nonempty ID matters for duplicate handling in a web data stream. Align those fields with the store’s order export before evaluating ad results.
Run one controlled test order
- Record the test time, browser, product, expected total, currency, and a unique test order ID. Use a safe payment or test environment where possible.
- Complete the order once. Check that the store created one order and sent one confirmation.
- Inspect the GA4 purchase event in a debug view. Compare transaction ID, value, currency, and item details with the order.
- Use Google’s Google Ads conversion diagnostics and Tag Assistant to check whether the intended conversion action is receiving a signal.
- Repeat with a failed payment or abandoned checkout. Neither should become a completed purchase.
That is a recommended test procedure; it is not a claim that a particular Megvi client has passed every step. Save the order ID and screenshots or debug output so another reviewer can reproduce the finding. Keep test transactions out of revenue reporting where the platform allows.
Reconcile three views without adding them together
| View | What it answers | Check when it disagrees |
|---|---|---|
| Store orders | What the business actually recorded | Payment status, refunds, time zone, test orders |
| GA4 ecommerce | What the site measurement observed | Missing or duplicate events, transaction IDs, consent, item values |
| Google Ads | What its conversion action attributes to ads | Selected action, attribution settings, counting, import or tag status |
Different attribution and reporting windows can create differences even when tags work. Investigate an implementation error when the same order is absent, duplicated, or materially wrong. Investigate attribution when the event is correct but the channels credit it differently. Report confirmed sales and advertising-attributed sales as separate figures.
Work through the most common faults
Purchase fires on page load: check whether the thank-you page can be refreshed or revisited and whether the event is tied to a real completed order. Value is missing or fixed: map the actual order total and currency, and inspect tax, delivery, discounts, and refunds according to the business reporting definition. Two conversion actions count the same order: decide which purchase action bidding should use, then check the others’ primary or secondary role. Too few conversions: widen the diagnostic period carefully and inspect click-to-purchase lag rather than declaring a winner from a few orders.
Megvi’s published Flower Center case describes an audit of Meta Pixel/CAPI, Google Ads conversion tags, GA4 event mapping, and UTMs before campaign changes. The case shows the type of cross-platform troubleshooting involved; it does not provide a universal uplift estimate. Our UAE Google versus Meta guide helps select the channel once the purchase signal is credible.
Make the next decision from verified data
After the test, record the conversion action used for bidding, the date it changed, and the order count confirmed by the store. Review cost per verified purchase alongside margin and repeat purchases where available. Use the ROI and ROAS calculator to model the break-even point, then compare actual orders with that scenario. If tracking remains unreliable, fix the order path before increasing automated bidding budgets.
For a measurement audit tied to campaign decisions, discuss Google Ads with Megvi and bring a recent order export, conversion-action list, and tag access.
