Start verifying Meta Pixel and CAPI with one test order whose payment status you know. Find its browser and server events, compare the event names, shared identifier, value and currency, then check what actually deduplicated. That gives you a specific next step: repair the sender, resolve an amount difference, or collect the missing evidence.
8 parts · 3 checkpoints
Show all 8 parts
- 01Why one order can produce two events
- 02Define the customer action before the event
- 03Identify the senders before choosing an implementation
- 04Follow one order through verification
- 05Check identity, amount and match quality separately
- 06Match the symptom to the first check, has a checkpoint
- 07Learn to read the first week's data
- 08Record the current findings for the next review, has a checkpoint
Key point Verify one purchase first
A connection is the starting point. Check the order's two event reports, deduplication, amount and consent scope separately.
Our classroom Shopify store sells insulated tumblers. A cup lists for $39.99; after its offer, classroom order TEST-1042 has a purchase value of $34.99 in USD to verify. These are example figures, with no live advertising results. A later exercise uses a separate order with a 10% discount, shipping and tax; do not combine those two orders.
You can follow the lesson without writing server code. By the end, you should be able to identify each sender, find the evidence and name one repair to test next. Inspecting a real setup requires the relevant viewing permissions. Changes to scripts, apps, data sharing or production configuration belong to an authorized owner.
Start with your current question
Why one order can produce two events
Where Pixel and CAPI send from
Meta Pixel runs in the website visitor's browser. It can report actions such as viewing a product, adding it to a cart or completing a purchase. Page loading, repeated scripts, browser blocking and consent state can affect what it sends.
The Conversions API, or CAPI, lets a platform or server send events to Meta. The sender might be a Shopify channel, a partner, server-side GTM, a custom backend or a CRM. CAPI and the browser Pixel can send in parallel. The server does not have to wait for the browser, and CAPI is not a fallback that starts only after a browser event goes missing.
When both routes report the same purchase, Meta needs to recognize one business action. Deduplication handles those repeated reports. It does not create or remove real orders in your store.
Schematic; your account's screen may differ
Original field relationship diagram. Neither route has to arrive first. The merge represents the result to verify, not a passing result for your store.
Identify the action before judging the timing
This lesson uses Meta's recommended event ID and event name method. The browser's eventID matches the server's event_id; the browser event name matches the server's event_name. Both reports must describe the same action and go to the same Pixel/dataset. Separate orders need separate identifiers, while retries of one action should retain its identifier.
Meta's current documentation also requires the two reports to arrive within 48 hours of the first receipt of that identifier. That is a limit, not a reason to delay sending. Compare the business action's event_time with send and receipt times to investigate latency. The alternative fbp/external_id method has order-dependent limitations, so do not apply this lesson's arrival-order rule to that implementation. Meta's deduplication documentation
Use the three views below to inspect what the page sent, what the server sent, and whether those reports belong to one action. Each view calls for different evidence.
Inspect the order from three views
Find the purchase report from the page. Identify its trigger, timing and sender, then inspect Purchase, eventID, value and currency.
- Keep this evidence
- Test page, consent state, browser parameters and trigger source; check whether the theme, apps or tags send more than once.
- What it establishes
- A browser event establishes that this route sent a report. You still need the server and deduplication evidence. A missing browser event is not a reason to replace the ad creative.
Find the purchase report from Shopify, a partner or your backend. Whether it arrives earlier or later, it should trace back to the same order reference.
- Keep this evidence
- Sender, response summary, event_name, event_id, event_time, action_source, value, currency and sending delay.
- What it establishes
- A successful server response does not establish that the browser report merged. Server transmission must also respect consent and permitted purposes.
Compare both reports for the order: matching names, shared identifiers and the same receiving destination. Then use current Test Events/Diagnostics and the order evidence to inspect the result.
- Keep this evidence
- Side-by-side parameters, receipt times and actual deduplication evidence. If a particular badge is absent, inspect the parameters and diagnostics before concluding that the event duplicated.
- What it establishes
- After identifying one action, verify its value and consent scope separately. Arrival order can help diagnose latency; it cannot establish a pass on its own.
Do not assign the same identifier to two different orders to make a report total look right. Also investigate repeated triggers, retries and old webhooks within each route. Browser/server deduplication does not replace sender and retry controls.
Define the customer action before the event
An event name describes something that happened on the website. Check the product page, cart, checkout and order first, then compare the name Meta received. Otherwise, a page refresh might be reported as an add-to-cart action.
| Event | Customer action | Check first |
|---|---|---|
| ViewContent | Viewing a product detail page | Product ID, page URL and content_ids identify the current cup |
| AddToCart | Actually adding the product to a cart | Button or cart-drawer trigger, quantity and product ID; a refresh is not an add |
| InitiateCheckout | Moving from cart into checkout | Trigger stage, item quantity and value; the implementation determines whether a server report exists |
| Purchase | Reporting a purchase as defined by the integration | Order reference, trigger timing, payment state, event_id, value, currency and deduplication |
Shopify's current channel documentation associates Purchase with completing a purchase and viewing the thank-you page. A Purchase event alone does not establish settled payment. We use a classroom order with a known payment state; in your store, record the current integration's trigger definition and compare it with the order's status. Shopify's data sharing and event documentation
Choose an event below to see what to look for on each route. Whether product views, cart actions or checkout actions have server counterparts depends on the implementation. You do not need to add a server copy of every event by hand.
Follow events through the purchase journey
The browser commonly reports the product page view. Check its URL, product ID and content_ids. A server counterpart depends on the platform and integration.
- Keep first
- The page and parameters for that visit. Confirm that they do not identify a collection page or an old product.
Add the item, then inspect the event triggered by the button or cart drawer. If an app sends a server counterpart, check the shared identifier and repeated triggers.
- Keep first
- Action time, product, quantity and trigger source. Do not substitute a page refresh for an add-to-cart test.
Move from the cart into checkout. Check the stage the event represents, quantity and value. If the platform supplies a server event too, identify its sender.
- Keep first
- That checkout action and its browser parameters, plus the corresponding order or session reference when a server report exists.
Find classroom order TEST-1042. Compare its browser thank-you page or platform Customer events report with the server report from Shopify, a partner or the backend.
- Keep first
- Order and payment state, both Purchase names and identifiers, value, currency, deduplication and consent evidence.
Even when all four event names appear, verify the actions, identities, amounts and senders they represent. For product identifiers and consistent event definitions, continue with event taxonomy and QA.
Identify the senders before choosing an implementation
Your store may already have several senders
A sender is the code, app or platform that generates and transmits an event. Assign a primary sender for each event on each route. One browser route and one server route can coexist; the problem is unexplained overlap.
| Where to inspect | What to inspect | Who confirms it |
|---|---|---|
| Theme and old checkout scripts | Old Pixel or thank-you scripts, repeated installs, triggers surviving a theme change | Theme or store owner |
| Shopify Customer events | Custom pixels, subscribed events and current consent settings | Store and tracking owner |
| Facebook & Instagram channel | Connected Pixel/dataset, data sharing level and associated business assets | Authorized channel owner |
| GTM/server-side GTM | Tags, triggers, variables, regenerated event IDs and container version | Tag or engineering owner |
| Third-party tracking apps | Event scope, advanced matching, overlap with other apps and exit procedure | App and privacy owner |
Record the routes to retain, candidates to retire, their owners and a rollback method. Avoid turning off all valid signals blindly or adding another unknown tool to fix duplication. When changing an app, theme or channel, compare the before-and-after state with a new order reference. Change one reproducible issue at a time.
Check data sharing and consent separately
Shopify's Standard, Enhanced and Maximum options describe data sharing settings, not verification scores. Its current documentation describes Standard as browser Pixel sharing; Enhanced and Maximum include CAPI. Check which options and connection your store actually has in the Facebook & Instagram channel settings, and record the Pixel/dataset.
A server request avoiding a browser blocker does not grant consent. Check Customer events, the cookie banner or consent tool, customer region, permitted data categories, opt-out state and privacy policy. Missing events or identifiers can reflect a permission boundary. Establish the reason before treating every omission as an installation fault, and do not default to Maximum.
Choose among the four routes according to what your team can maintain. Your selection feeds the verification record later in this lesson. It changes this classroom record only, with no change to Shopify or Meta.
Choose a route you can explain and maintain
Suits a store starting with its current Facebook & Instagram channel and data sharing settings.
- Browser
- Record the current Pixel, sharing level, Purchase page and trigger stage.
- Server
- If the selected sharing setting supplies CAPI, establish which dataset receives the order event. A connection badge is not event evidence.
- Migration risk
- Old theme code, apps or GTM may still send. Inventory them first.
- Test sequence
- Record the settings and Pixel ID, place a newly referenced test order, then compare both routes, deduplication and value.
- Consent scope
- Check the privacy notice, customer choices and approved purposes. A classroom selection cannot stand in for customer consent.
Suits a team with clear browser and server ownership that needs one account of its fields, changes and verification.
- Browser
- Identify the primary route for this event in Customer events, GTM or the theme, including the Purchase trigger.
- Server
- Identify the sender, order reference, event_time, event_id, value, currency and response summary.
- Migration risk
- Without a shared identifier, the routes may duplicate a purchase. Hold new senders while investigating; do not indiscriminately switch off existing tracking.
- Test sequence
- Keep both parameter sets, repair one reproducible issue, then retest with a new order reference.
- Consent scope
- Check consent and purpose for each route. Server transmission does not confer additional permission.
Suits a team using a supported partner or CAPI Gateway, with named people who can inspect events and migration state.
- Browser
- Identify the primary sender so the partner and an old theme script do not report overlapping Purchases.
- Server
- Record event scope, dataset, accessible error evidence, partner contact role and internal owner.
- Migration risk
- An old route left running, an unverified new route or separately generated IDs can leave duplicates.
- Test sequence
- Record the old route's end date and exit procedure, then compare both reports for a newly referenced test order.
- Consent scope
- Check the contract, consent and privacy notice. A partner connection cannot expand the permitted purpose.
Suits a team with an engineering owner, reliable order source and safe release process. The field examples below are insufficient for a production deployment.
- Browser
- Identify the primary sender and explain how its eventID maps to the same order.
- Server
- Record the order source, event generation point, idempotency and retry rules, event_id, log summary and controlled-test owner.
- Migration risk
- Retries, old webhooks and parallel code must not become separate purchases. Record the changed version, rollback method and old route's state.
- Test sequence
- Trace how one action generates an event, compare browser/server fields, then have authorized engineers arrange a controlled retest.
- Consent scope
- Keep non-sensitive summaries here, without tokens, customer data or full logs. Review the approved scope before release.
If you cannot yet identify the senders, complete the inventory before deciding whether you need another tracking service. When asset ownership or viewing permissions are unclear, review accounts and assets first.
Follow one order through verification
Start with the Shopify order
The test order gives both routes a common reference. Keep its reference, payment time and status, items, offer, tax, shipping, currency and refund state before looking for events. That reduces the risk of matching reports from two purchases made close together.
The sequence below allocates about 30 minutes to a classroom review. It helps assign the work; these are not platform processing deadlines. If evidence is missing, record the uncertainty rather than passing the order to finish on time.
Five stages for one order
0–5 minutes
Establish the order facts
Where in the UI: ShopifyOrdersTest order
Find TEST-1042 and record its payment state, time, items and order value. Distinguish list price, discounted item value and order total. Check whether the test has been refunded.
Keep as evidence
Keep a redacted order summary that another reviewer can locate. Do not enter customer names, email addresses or payment details here.5–12 minutes
Find the browser report
Where in the UI: Meta Events ManagerCurrent datasetTest Events
In the authorized dataset, locate the browser Purchase by action time. Check the trigger page, event name, eventID, value, currency, product identifier and consent state.
If it goes wrong
If it is missing, inspect consent, page loading, blocking and the primary sender. A server event arriving first does not replace this check.12–20 minutes
Find the server report
Where in the UI: Channel, partner or server event evidence
Locate the CAPI report using the same order reference. Compare event_name, event_id, event_time, action_source, website source URL, amount, currency and sending result. Distinguish action time, send time and Meta receipt time.
Keep as evidence
Keep the sender, response/error summary, delay and corresponding order reference. Do not copy tokens, full customer data or raw production logs into the lesson record.20–27 minutes
Compare deduplication and value
Where in the UI: Test EventsDiagnostics and Shopify order
Establish that both reports describe one action, then inspect the deduplication and diagnostics available in the current interface. Separately reconcile value, currency, discount, tax and shipping definitions, and explain refund treatment.
How to confirm
Seeing both routes establishes two reports. Keep evidence for their identifiers, observed deduplication result and order amount.27–30 minutes
Name one repair to retest
Record the unresolved issue, owner, evidence location, review date and one specific action. Retest a change with a new order reference so the next result can explain that change.
You should see
You have a record an authorized person can review. While identity, amount or consent remains unresolved, do not use these readings to change bidding, start value optimization or increase budget.
Four places answer four questions
Labels vary by account and version. Confirm the dataset and store first, then use the question below to find the relevant view. One screenshot cannot establish everything.
user_data contains customer information parameters for matching. It is not an instruction to send everything you can collect. Even fields that require hashing must first satisfy consent and purpose requirements; hashing does not grant collection or sharing permission.
Check identity, amount and match quality separately
How the shared identifier appears on each route
These excerpts show relationships for the classroom order only. In the teaching product code test-20oz-blue, 20oz labels capacity and blue labels color. The excerpts are incomplete examples. An engineer must check the placeholders, required parameters and current API version separately; neither excerpt is ready to deploy.
The browser Pixel passes eventID in the fourth argument:
fbq('track', 'Purchase', {
value: 34.99,
currency: 'USD',
contents: [{ id: 'test-20oz-blue', quantity: 1 }],
}, { eventID: 'test-order-1042' });
The server uses event_id. Time, website URL and the matching field below are deliberately placeholders:
{
"data": [{
"event_name": "Purchase",
"event_time": "<unix-time-for-test-order>",
"event_id": "test-order-1042",
"action_source": "website",
"event_source_url": "<test-checkout-url>",
"user_data": { "em": "<hashed-after-approved-consent>" },
"custom_data": {
"currency": "USD",
"value": 34.99,
"contents": [{ "id": "test-20oz-blue", "quantity": 1 }],
"order_id": "TEST-1042"
}
}]
}
event_name and the shared identifier identify an action. event_time describes when it happened, action_source describes its source, and event_source_url identifies the website location. order_id helps reconcile the business order; it does not automatically replace the implementation's agreed event_id. Different events within one order also need a naming rule that distinguishes their actions.
Why the amount might not match
value is the amount reported under the integration's definition; currency identifies its currency. The list price and the classroom purchase value are $5.00 apart, 12.5 percent, and this lesson does not explain that gap for you: find which discount, voucher, or shipping rule produced it in the offer record, then confirm both event routes use one amount definition and one currency. Both event routes should use consistent amount definitions and currency.
Shopify's current documentation describes total order price, including products, duties, taxes and discounts. Still inspect what the current integration sends: when the discount is applied, how shipping is included, whether tax is inclusive or additional, how refunds affect later records and whether currency conversion occurs. Do not read every value as net revenue or profit.
What EMQ and attribution tell you
Two terms to distinguish when reading scores and reports
- Event Match Quality (EMQ)
Helps diagnose whether matching parameters can associate an event with a Meta user. Available email, phone, IP, browser or click identifiers can affect it, as can consent.
If misread EMQ is neither a deduplication result nor a profit score. Do not collect beyond the approved scope to improve it.- Attribution
Assigns a conversion to an ad or channel under particular rules and windows. Its purpose differs from the Shopify order ledger.
If misread Even with correct orders, values and events, Meta, Shopify and GA4 can differ because of windows, cross-device behavior, refunds, payment state or consent. Explain the gap before judging tracking broken.
Match the symptom to the first check
Identify which evidence is missing
“The purchase count is wrong” can describe several different problems. The options below suggest possible causes and the first evidence to collect. They do not read your account or automatically identify a faulty script.
Choose the symptom you can observe
Trace the browser route for this test order first.
- Possible cause
- Page loading, consent, a blocker, a theme change or old code.
- First check
- Follow the order's page and action time to the primary browser sender before replacing ad creative.
- First evidence
- Page URL, browser report or missing-event summary, consent state, and theme/app change history.
- Hold for now
- Do not use these conversion readings to increase budget while the browser route remains unexplained.
Find the server's sending record to distinguish an event that was never sent from one that arrived late.
- Possible cause
- Integration authorization, containers, queues, batching or server event fields.
- First check
- An authorized owner compares send times, responses and error summaries with event_time and action_source.
- First evidence
- Server sender, order reference, response code, action time and send/receipt delay. Follow the same route if the server event is absent.
- Hold for now
- Do not stack another app on top to supply a missing report before checking the original route.
Establish whether the two reports on screen describe one action.
- Possible cause
- Different or regenerated IDs, different receiving datasets, or extra senders.
- First check
- Find both parameter sets and current deduplication diagnostics for the order rather than counting rows on screen.
- First evidence
- The order's event_name, eventID/event_id, receiving destination, times and deduplication evidence.
- Hold for now
- Do not use a potentially inflated purchase count to judge ROAS or increase budget.
Explain the difference between order and event amounts before treating a value change as better advertising performance.
- Possible cause
- Discounts, tax, shipping, refunds, multiple currencies, each route's amount definition, product mix or high-value orders.
- First check
- Compare the order's amount components, currency and both routes' definitions. Sample by order-value band when needed.
- First evidence
- Redacted order amount breakdown and both reports' value, currency and item quantity.
- Hold for now
- Do not change bidding or campaign objectives, or start value optimization, while the amount remains unexplained.
Practice the repair order with the cup store
Choose a first action for each scenario. The first uses TEST-1042. The second is a separate discounted order; its discount rate does not explain the earlier $34.99 value. Feedback teaches an investigation sequence without operating the store.
What would you inspect first?
Shopify has one TEST-1042 with classroom value $34.99. Events Manager has browser and server Purchase reports, without enough evidence yet to establish deduplication.
- First evidence
- Same order reference, Test Events summary, browser/server parameters and both event identifiers.
A separate cup order uses a 10% discount and includes shipping and tax. Shopify, browser and server amounts differ, and the difference is unexplained.
- First evidence
- Redacted Shopify breakdown, browser purchase parameters, CAPI custom_data value/currency, and discount/tax notes.
The store retains an old theme Pixel, channel integration, Customer events and GTM. Nobody can explain which ones send Purchase.
- First evidence
- Theme, Customer events, Facebook & Instagram channel, GTM, server container and third-party app inventory.
The store changed Business Portfolio. Its channel says connected, but old and new datasets are mixed and neither the sharing level nor current consent scope is recorded.
- First evidence
- Non-sensitive summary of current channel/sharing settings, Pixel/dataset identifiers and Business Portfolio.
Each scenario points to a different first check: identity, amount, trigger source or asset connection. Work on the issue supported by evidence, then use a new order to inspect the change so the next reviewer can interpret the result.
Learn to read the first week's data
Expand from one test to representative orders
One order can expose a specific fault; it cannot establish that a whole week or every customer works correctly. Start the first week by inspecting 10 real orders. Inspect all of them if there are fewer; with more orders, cover markets, devices, payment methods, discounts and order-value bands. This is a troubleshooting sample, not a statistical significance or delivery-stability threshold.
For each order, keep the reference, payment time, items, value, currency, discount and refund state. Compare browser/server names, identifiers, action times, senders and consent. Check whether permitted matching parameters and EMQ differ with the traffic mix; do not expect identical fields for every customer.
Either route may arrive first. For a delay, compare action time, queue/send records and receipt time. For changing order values, inspect high-value products, currencies and refunds first. Keep explanations for understood Meta/Shopify/GA4 attribution differences. Escalate unexplained widening gaps, conflicting IDs or amounts, and unexplained UTM issues to the relevant owner.
Correct events do not establish incremental ad sales
Deduplication asks whether an action was reported twice. Reconciliation asks how amounts correspond. Attribution asks which ad receives credit under a rule. Incremental advertising effect asks a further question: how many purchases would have happened without the ad?
Gordon and colleagues' 2019 study compared US Facebook advertising experiments with observational estimates. Even after extensive demographic and behavioral adjustment, observational methods often did not reproduce randomized results. This older study helps distinguish measurement methods; it did not test this store's CAPI and does not predict a current performance lift. Verified tracking does not make every attributed purchase an additional purchase caused by ads. [1]
Retest after changes to themes, apps, data sharing, checkout, offers or currency, and when an order anomaly remains unexplained. For ongoing retries, rollback, monitoring and incident ownership, continue with CAPI governance. Here, finish the path from one order to one reviewable record.
Record the current findings for the next review
What the record should retain
A useful record lets another person locate the same order, identify both senders, see what remains uncertain and know who will retest. The form retains the implementation route, sharing state, browser/server senders, migration stage, shared identity, consent scope, test order reference, evidence location, owner, date, next action and repair sequence. Each helps reproduce the check or make the next decision; “connected” alone cannot do that.
The evidence location should point to the four events' actions, senders and deduplication comparison, the order's amount and payment summary, and explanations of consent and cross-platform attribution differences. Assign owners for Pixel/CAPI, sharing settings, server parameters and order reconciliation. One person may cover several roles, but every check needs an owner.
The record stays in this browser. Save it manually to restore, export or clear it; you can also import a record from the earlier cup-store lesson. Do not enter access tokens, passwords, card details, customer data or raw payloads. An evidence location can be a non-sensitive description of a restricted team folder.
Your earlier route selection is included above. The decision below identifies the first issue to address in this record. Completing a form does not establish that live events, values or privacy settings passed.
What needs attention first?
Fill the missing sender, stage, order evidence or review details before judging the repair sequence. Shopify data sharing state, Primary browser sender, Primary server sender, Migration stage, Browser/server identifiers for this action, Consent and privacy scope, and 6 more
An unknown browser or server source leaves this Purchase unexplained. Complete the inventory before arranging a change, and hold budget increases based on these readings.
Record when the old route ends, what stays and how to roll back. Have an authorized owner make a controlled change, then retest a new order reference.
Both reports of this action need matching names and eventID/event_id. Inspect real parameters and deduplication evidence before judging the signal usable.
Check Shopify's current settings, customer choices and approved purposes. Server transmission cannot replace this review; hold relevant audience activation while scope is unclear.
Adding a tool or changing several routes cannot explain the current anomaly. Return to the order evidence and name one issue with an owner and a retest.
These fields support a review appointment. Verify actual deduplication, amount and consent in the current systems. This record executes no configuration and does not establish readiness to scale.
Decide from evidence whether to continue
These are operational conditions, independent of the form's status color. Meet the relevant condition before moving to the next check.
Use evidence to choose the next step
| Done when | Stop if |
|---|---|
Both reports, the receiving destination and current diagnostics explain the deduplication result for one action. Continue to the amount check. | The shared identifier or actual deduplication remains unclear. Hold ROAS conclusions and budget increases based on this purchase count. |
| Evidence Browser/server parameters, times and deduplication evidence for the same test order. | |
Discount, tax, shipping, refund and currency definitions reconcile item by item. Use those explained amounts for further analysis. | Value/currency still differs from the order without an explanation. Hold value optimization. |
| Evidence A redacted order breakdown and both routes' value/currency comparison. | |
The owner confirms retained routes, current dataset and approved scope. Arrange one controlled change and a new test order. | Sender, asset ownership or consent scope is unclear. Hold new sending routes and relevant audience activation. |
| Evidence Sender inventory, setting/consent summary, owner and review date. | |
Product view, add-to-cart, checkout and purchase actions, expected routes and fields are checked. Explain and retain any differences before continuing to event taxonomy and QA. | A core event's action or sender is unexplained, or cross-platform differences remain unresolved. Do not treat this record as a complete verification. |
| Evidence Four-event check, order comparison, and screenshots or redacted summaries from Meta/Shopify/GA4 if installed, with explanations of differences. | |
Two Purchase reports for one order: what comes first?
Self-check not answered
Explain it in your own words: who sent this order, why do both reports represent one action, what explains the amount, and what one issue will you repair next? Put any unanswered part in the record for the next evidence review.
Research
- Gordon et al., A Comparison of Approaches to Advertising Measurement: Evidence from Big Field Experiments at Facebook, Marketing Science, 2019, 38(2). Qualitative finding from the institutional abstract. Historical US Facebook experiments, with Facebook research coauthors; not a test of current CAPI performance.
No checkpoint finished in this lesson yet. Start with “Go to the exercise”.
Self-check not answered
Common questions
- What is the difference between Meta Pixel and CAPI?
Pixel sends from a website visitor's browser; CAPI sends from a platform or server. They can report one business action. Verify the shared event identifier, name and deduplication, then check amount and consent separately.
- Why begin with one test order?
It supplies a known product, payment state, amount, currency and time for comparing routes and senders. One passing order helps resolve specific faults; it does not establish that all traffic works correctly.
- Does a server event arriving first mean tracking is broken?
Arrival order alone cannot judge the shared-ID method taught here. Check names, IDs, receiving destination, receipt times and actual deduplication. Meta currently requires both reports within 48 hours of the first receipt of that ID; other methods can have different limitations.
- What should I check first when one order has two Purchase reports?
Establish whether browser and server report one action, compare names and shared IDs, then inspect deduplication and repeated senders. Two rows are neither automatically two orders nor proof that a merge occurred.
- Can I increase budget after EMQ improves?
EMQ diagnoses matching parameters. It does not establish deduplication, correct amounts, profit or incrementality. Verify events and value, then judge budget against your business goals, costs and observation window.
- What should I inspect in Shopify data sharing?
Record the level, connected Pixel/dataset, Customer events and other senders. Check region, consent categories, opt-out state and privacy notice. Do not default to Maximum or treat server transmission as additional data permission.
- Which CAPI field relationships should I inspect?
Compare event_name, event_id, event_time, action_source, website source URL, permitted matching parameters, value/currency and product identifiers. Then check response, delay, actual deduplication and the order. A field example cannot replace current-version parameter checks.
- What do the practice and record establish?
Practice helps choose an investigation sequence; the record retains implementation, senders, order evidence, consent and review plans. Neither changes an account nor proves live tracking passed. An authorized person still needs system evidence. The section “What the record should retain” describes the complete record.
What to study next
Choose by the issue that remains unresolved, keeping this test order's record for the next review.