Conversion Tracking
Part of Paid social measurement
Comparing platform-reported results with first-party sales
Reconcile attributed purchases with confirmed orders by defining each population and checking event, date, value and order-status rules.
Put the ad platform’s attributed purchases beside confirmed orders defined from your own records. Keep both figures and explain their populations. A platform count applies its event and attribution rules; an order count applies the business’s payment, cancellation and refund rules. Equal totals would not prove that the same orders were counted or that the ads caused them.
Define each population
In TikTok Ads Manager, open Analytics > Attribution Analytics; Overview is the default page. Its Performance Overview covers purchases attributed to TikTok campaigns and includes CPA and Total Spend. Record the exact purchase metric and value labels displayed in your account rather than assuming a column name.
Use Performance Comparison in TikTok Attribution Analytics to compare click-through and view-through reporting across available attribution windows. Record the selected windows, event, dates, time zone and currency. For Meta Ads Manager or another platform, record that platform’s own displayed purchase metric and settings rather than borrowing TikTok terminology.
TikTok says the optimal view-through attribution (VTA) window depends on business needs and recommends 24 hours/1 day to understand conversions within 24 hours of ad impressions. It says most measurement partners provide VTA windows of 1–24 hours. Record the actual click-through and view-through windows used; do not assume the recommendation is the setting applied to your account.
Define a confirmed order before extracting first-party data. Decide whether payment must clear, whether unfulfilled orders count, and how cancellations, refunds, exchanges and test records are treated. Count distinct eligible order IDs and calculate their value on a stated basis.
Shopify’s Sales by Vendor report lists gross sales, discounts, returns, net sales, tax and total sales. Use those fields to inspect value composition, not as a substitute for defining eligible orders from first-party order records. A Shopify sales-report total is not automatically this population; check report definitions and settings against your eligibility rules.
For an Australian comparison, state whether each value includes GST, shipping and discounts. Shopify’s Sales by Vendor fields distinguish tax from gross and net sales, and include tax in total sales; verify the platform figure’s value basis in its own reporting before comparing.
Show the store-wide confirmed-order total as business context. If campaign tags or another permitted connection identify a subset of orders, show that subset separately and describe how it was selected. An untagged order may still have followed an ad interaction; a tagged order is not necessarily incremental. Do not label all store sales as campaign sales.
| Field | Platform report | First-party record |
|---|---|---|
| Population | Attributed events for selected campaigns | All eligible orders, or a separately defined traceable subset |
| Date | Report setting and attribution window | Chosen order, payment or adjustment date |
| Count | Reported purchase events | Distinct eligible order IDs |
| Value | Reported event value and currency | Eligible order value on a stated tax, shipping and discount basis |
| Later changes | Updated platform reporting | Refunds, cancellations and edits under the business rule |
Platform-Reported Purchases vs First-Party Confirmed Orders
- Population
- Attributed events for selected campaigns (TikTok Ads Manager)
- Population
- All eligible orders, or a separately defined traceable subset (first-party records)
- Date
- Report setting and attribution window (TikTok)
- Date
- Chosen order, payment or adjustment date (first-party)
- Count
- Reported purchase events (platform)
- Count
- Distinct eligible order IDs (first-party)
- Value
- Reported event value and currency (platform)
- Value
- Eligible order value on a stated tax, shipping and discount basis (first-party)
- Later changes
- Updated platform reporting (e.g. post-click refunds)
- Later changes
- Refunds, cancellations and edits under the business rule (first-party)
Investigate the difference
First check whether the platform event represents a completed purchase. A checkout start, repeated thank-you page load or duplicate event may explain an apparent sale without another order. Flag the affected dates for a tracking investigation; detailed event repair belongs in conversion tracking.
Check the attribution windows and view-through treatment before interpreting the gap. TikTok can attribute a conversion after a view without a click, and its Attribution Analytics compares click-through and view-through results across windows. TikTok says VTA conversions are commonly higher than click-through attribution conversions because VTA is based on an impression and click-through attribution on a click.
Next compare dates and values. An ad interaction and purchase can fall on different days, while a refund can be recorded later. Tax, shipping and discounts may be represented differently. State the basis of any internal comparison instead of relabelling either system’s metric.
Where a lawful, usable connection exists, inspect a small set of event and order identifiers to investigate specific gaps. Aggregate totals alone cannot identify which orders overlap. Keep personal information out of the review unless it is necessary and permitted.
Record the decision
Keep a dated variance log for each comparison. Record the period, platform and displayed metric, attribution settings, first-party eligibility rule, count and value from each population, differences checked, what is explained, what remains unknown, and which business figure governs the decision.
Do not add purchases attributed by several platforms into a total-sales figure; the same order may be eligible for credit in more than one system. A close match to first-party sales also does not establish additional sales. Use a suitable comparison design when that causal question matters.
Reconciling Ad Platform Attributions with First-Party Sales
- Define populations clearlySpecify what counts as an attributed purchase and a confirmed order (including GST, shipping, refunds)
- Record settings and metricsCapture attribution windows, date ranges, time zones, currencies and source report names
- Investigate discrepanciesCheck for tracking issues, timing mismatches, and differences in data rules
- Log variances and decisionsMaintain a dated variance log with explanations and unresolved items
- Avoid double-countingDo not sum purchases across multiple platforms if same order may be counted twice
