Shopify: 3 months for $1/month, plus up to $10,000 credits as you sell
Tutorial Series/Google Ads Basics
Intermediate55 minutesStep 11

Enhanced Conversions and Value Quality Check

Use an enhanced conversions and value QA sheet, value QA admin evidence paths, official evidence boundaries, 20oz value trust lab, and value fault router to validate Enhanced conversions / Diagnostics, Google tag / GTM, Consent Mode, transaction ID, purchase value, currency, GA4 purchase, Shopify Orders / Refunds, post-refund revenue, SKU / GTIN / product group, Bid strategy report, Change history, and Target ROAS readiness.

11
Current Lesson
11/11 lessons
Reviewed by Ranfeng Wei. Maintained monthly against Shopify, Google Search, ads, analytics, and ecommerce operating workflows.
Quick Answers

TL;DR: Put hashed first-party data, field source, consent status, transaction ID, purchase value, currency, post-refund revenue, diagnostic severit

Q: What is the key action in this lesson?A: Put Google Ads value, GA4 purchase revenue, Shopify gross / net sales, discounts, tax, shipping, refunds, chargebacks, currency, and product

Lesson Progress
Progress
11/11 lessons
Current lesson unlockedContinue in sequence

Lesson HowTo steps

Complete this lesson in 4 steps

  1. 1

    Write the enhanced conversions and value QA sheet first

    Put hashed first-party data, field source, consent status, transaction ID, purchase value, currency, post-refund revenue, diagnostic severity, and tROAS trust level on one sheet. Do not treat enhanced conversions as proof that value is trustworthy.

  2. 2

    Reconcile the latest 20 orders row by row

    Put Google Ads value, GA4 purchase revenue, Shopify gross / net sales, discounts, tax, shipping, refunds, chargebacks, currency, and product group on one row. Confirm the differences are explainable before value enters automation.

  3. 3

    Use the 20oz value trust lab to choose the action

    For the 20oz tumbler discount order, Canada currency drift, duplicate thank-you-page purchase, and high-refund SKU, choose fix value definition, fix currency, fix transaction ID dedupe, or add post-refund review. Do not default to raising tROAS trust.

  4. 4

    Write the tROAS stop or go rule

    Use the value fault router and trust levels to mark the current state as stop, low trust, observation trust, or testable trust. The transfer record should include the problem, first evidence, repair action, responsible lead, review window, and rollback line.

Article FAQ

Answer the common misunderstandings first

If enhanced conversions are working, can I hand value to tROAS?

Not automatically. Enhanced conversions improve matching stability for existing conversions. They do not define purchase value, currency, post-refund revenue, margin, or transaction ID deduplication. If value is wrong, tROAS learns the wrong amount more consistently.

Why sample the latest 20 orders row by row?

The risky state is not only missing value. It is Ads value that looks normal but does not reconcile to GA4, Shopify net sales, discounts, tax, shipping, refunds, chargebacks, or currency. A 20-order sample exposes duplicate purchase, mixed currency, pre-refund gross revenue, and pre-discount value.

What does the 20oz value trust lab teach?

It trains you not to raise tROAS trust just because ROAS improves. First decide whether the fault is value definition, currency, transaction ID dedupe, or post-refund review. Each fault needs a different repair action and a written QA-sheet line.

What should I have after finishing this lesson?

You should have an enhanced conversions and value QA sheet covering field source, consent status, transaction ID, value, currency, post-refund revenue, 20-order reconciliation, diagnostic severity, tROAS trust level, responsible lead, review window, and rollback line.

What are enhanced conversions, and can they replace conversion tracking?

Enhanced conversions use hashed first-party data on top of existing conversion tracking to help Google match conversions more reliably. They are not a new purchase event, and they do not replace the Google tag / GTM, Consent Mode, transaction ID, value, or currency. If the base event is wrong, enhanced conversions only make the wrong signal more consistently matched.

Should purchase value send gross sales, net sales, or profit?

Decide the business definition first. Many ecommerce stores send order revenue first, then reconcile refunds, discounts, tax, shipping, and margin in the review sheet. If you want to feed profit or margin into automated bidding, each order needs a stable field source. Do not mix gross sales one week and net sales the next, or tROAS will learn a blended definition.

What should I do when Google Ads conversion value does not match Shopify sales?

Do not change bidding first. Sample the latest 20 orders and compare transaction ID, purchase time, currency, discount, tax, shipping, refund, duplicate purchase, and attribution window. Small differences are normal. Large differences should be traced to tracking, attribution, refunds, currency, or duplicate events before trusting tROAS.

When can tROAS trust my value data?

At minimum, the primary conversion action must be correct, value / currency / transaction ID must reconcile to order samples, and refunds or low-margin SKUs must not mislead the system for long periods. Then observe for 4 weeks or 3 conversion cycles. If the value definition is still changing, do not use tROAS as an amplifier.

Loading interactive version
Text version of this lessonExpand

Google Ads automation depends on conversion and value signals. Many accounts do not fail because bidding tools are missing. They fail because purchase value, currency, transaction ID, refunds, consent status, or enhanced-conversion fields are not trustworthy. Enhanced conversions can improve matching quality, but they do not define business value for you. This lesson gives you an enhanced conversions and value QA sheet plus a value fault router for deciding what can be trusted by tROAS or Maximize conversion value.

Lesson output: enhanced conversions and value QA sheet

This sheet puts technical implementation and business value into the same evidence table. Google Ads can learn from the conversion value you send. If that value is pre-discount revenue, mixed currency, duplicate orders, or pre-refund gross revenue, automation learns the wrong target more consistently.

Check layer Inspect first Pass condition If failed
Match fields Where email, phone, name, and address come from; whether they appear reliably and respect consent Field source, format, and consent status are explainable Do not upload dirty fields or bypass user choice
Deduplication Transaction ID, order ID, duplicate purchase, confirmation refresh, and multiple tags One order counts once and reconciles across Ads / GA4 / Shopify Do not trust value automation while duplicate orders remain
Value definition Product revenue, tax, shipping, discounts, pre-refund / post-refund revenue have fixed rules Purchase value and business review definitions can explain differences Do not switch to aggressive tROAS before value definition is fixed
Currency and market Currency, market currency, conversion logic, and Google Ads billing currency Cross-market values can be compared and currency code is stable Do not make cross-market scale calls while currency drifts
Privacy and consent Consent Mode, cookie banner, privacy policy, tag behavior, and data-use explanation User choice and platform sending behavior are aligned Do not sacrifice consent boundaries for match rate

Define the terms before using the sheet

Enhanced conversions use hashed first-party customer data to improve conversion matching. They are not a new conversion event. Email, phone, name, or address fields can help Google match an existing conversion back to an ad interaction, but they cannot fix a wrong purchase event, duplicate order, bad value, bad currency, or refund definition.

Hashed first-party data is customer information you collected directly, transformed by SHA256 or a similar one-way hashing method so the original text is not directly readable. Source, formatting, and consent boundaries still need to be explainable.

Transaction ID is the unique order or lead identifier used to help prevent duplicate conversion counting. Without a stable transaction ID, confirmation-page refreshes, multiple tags, or duplicate scripts can count one order more than once.

Purchase value is the order value sent to the ad system. You need to define whether it includes product revenue, tax, shipping, discounts, refunds, and canceled orders. If value is wrong, automation becomes more consistently wrong.

Currency is the currency code attached to value, usually an ISO 4217 three-letter code such as USD, CAD, or EUR. Cross-market accounts can easily distort ROAS, tROAS, and budget decisions when currency is missing or mixed.

Consent status is the user choice about advertising, analytics, or personalization data use. Tags and platforms adjust collection, modeling, and sending behavior based on consent state. Match rate should not override user choice.

Consent Mode is Google's mechanism for adjusting tag collection and modeling behavior according to user consent state. You see it around the cookie banner, Google tag, GTM, and privacy settings. If consent state and tag behavior disagree, enhanced-conversion match quality and privacy boundaries both become unreliable.

checkout is the path where a shopper moves from cart to payment, address entry, shipping selection, and completed purchase. It is where purchase value, transaction ID, customer fields, and consent state often meet. If checkout fields are missing or fire twice, value QA becomes dirty at the source.

Enhanced conversions solve recognition, not business definition

Layer Solves Does not solve
Enhanced conversions Improves the stability of recognizing and matching existing conversions Does not define net revenue, refunds, margin, new-customer weight, or low-quality orders
transaction ID Helps one order avoid duplicate counting Does not guarantee order value, currency, or refund definition is correct
value-based bidding Lets the system optimize toward conversion value or ROAS goals Does not know margin, cash recovery, or refund risk unless value definition is designed clearly

Official evidence boundary: platform facts versus business definitions

Enhanced conversions, Consent Mode, conversion values, and Target ROAS are connected, but they are not the same job. Official Google documentation can tell you what the platform can collect, hash, match, model, and optimize against. It cannot decide your business definition of net revenue, post-refund revenue, margin, new-customer value, cash-flow risk, or which orders you are willing to teach automation to chase.

Enhanced conversions supplement existing conversion tracking by sending hashed first-party customer data, such as email, phone, name, or address fields, from website tags or imported offline events. That is a measurement recognition layer. It is not a new purchase event, and it does not repair a bad value definition, duplicate order, missing currency, or refund policy.

Google tag setup adds another boundary. You still need to know which conversion page contains user-provided data, whether customer data terms are accepted, which collection method is used, and whether consent parameters allow the tag to send user-provided data. If Consent Mode is implemented, ads-cookie behavior is subject to ad_storage. That means a clean QA note should say both: "the platform can send this field under this consent state" and "the business still owns the cookie banner, privacy explanation, and user choice."

Conversion values and Target ROAS have the same separation. Google Ads can use conversion values for ROI review, Maximize conversion value, and Target ROAS. But the account owner must decide whether the value means product revenue, gross revenue, net sales, profit, pre-refund revenue, or post-refund revenue. For Search campaigns that newly started reporting values or changed value reporting, the safer move is to include the new values in the Conversions column and wait 4 weeks or 3 conversion cycles before trusting Target ROAS decisions.

Official layer Can prove Cannot prove
Enhanced conversions Hashed first-party data can supplement existing conversion matching from website tags or imported offline events. Order value, currency, post-refund revenue, margin, or new-customer weighting is correct.
Google tag + Consent Mode User-provided data field source, customer data terms, and ad_storage consent status can explain whether data should be sent. Your cookie banner is compliant, your privacy explanation is clear, or user choice can be bypassed for match rate.
Conversion values Conversion value can support ROI review, Target ROAS, and Maximize conversion value bidding. The value definition should be gross revenue, net revenue, profit, or post-refund revenue.
Target ROAS readiness Campaign-type thresholds and the Search waiting window after value changes can be checked. Your first ROAS target is realistic or low-quality value will become profitable after automation scales it.

Enhanced conversions passing only means the matching path may be stronger. It does not prove the value definition, currency, refunds, margin, or tROAS target is trustworthy. A useful value QA record lets the ad lead, developer, operator, and finance reviewer see the same fields: which fields come from Google tag / GTM, which fields come from Shopify orders, which fields come from GA4 purchase, which fields come from the finance net revenue sheet, and which fields decide whether value can be handed to automation this week.

Evidence path Admin surface Fields to record Proves / still cannot prove Release rule / blocked move
Enhanced conversions matching evidence Google Ads -> Goals -> Conversions -> conversion action -> Enhanced conversions / Diagnostics; Google tag or GTM; Shopify checkout / customer fields. conversion action name, primary / secondary, tag source, enhanced conversions status, diagnostics severity, email / phone / address field source, pre-hash format, SHA256 / auto collection, match quality note, checkout path, test order id, last checked time. Proves whether enhanced conversions have stable first-party field sources; still cannot prove purchase value, currency, refund, margin, or the tROAS target is trustworthy. Release matching quality only; do not treat enhanced conversions status as tROAS release.
Consent Mode and user-data consent boundary Google tag / GTM Consent settings; CMP / cookie banner; Google Ads diagnostics; privacy policy and checkout consent copy. ad_storage, ad_user_data, ad_personalization, analytics_storage, default / update timing, region rule, banner choice, user data terms accepted, privacy policy URL, consent test order id, whether enhanced conversions fields send under granted / denied paths. Proves whether user data is used under the right consent state; still cannot prove value amount is correct or that legal copy covers every market. Expand enhanced conversions, audiences, or remarketing only after consent boundaries are clear; do not change denied to granted to raise match rate.
Order value and currency reconciliation Google Ads Conversions / Segments; GA4 purchase parameters; Shopify Orders / Timeline / Refunds; finance net revenue sheet. order id / transaction_id, Ads value, GA4 revenue, Shopify gross sales, net sales, tax, shipping, discount, currency, market, payment currency, converted value, refund amount, chargeback, post-refund revenue, value definition. Proves whether the amount Google Ads learns can be explained row by row; still cannot prove high-value orders are high-profit. Raise the value gate to observation or testable trust only after the 20-order reconciliation passes; do not switch to Maximize conversion value while currency is missing, purchase duplicates, or refunds are not reviewed.
Refund, margin, and SKU value quality Shopify Orders / Refunds; product margin sheet; Merchant Center product id / GTIN; Google Ads product group / PMax listing group. SKU / variant id, product id, GTIN, COGS, shipping cost, payment fee, refund reason, return window, pre-refund value, post-refund revenue, contribution profit, refund rate, chargeback rate, support cost tag, excluded SKU, next review date. Proves whether high value comes from real high-contribution products or from high-refund, low-margin, high-support-cost SKUs; still cannot prove matching fields are correct. Hand only SKUs that pass refund, margin, and support-cost checks to value-based automation; do not let high-refund SKUs keep training tROAS at full gross value.
tROAS release and rollback record Google Ads Bid strategy report; Change history; Budget / Target simulators; internal value QA release record. current bid strategy, target ROAS, conversion value source, Conversions column inclusion, observation window, 20-order sample pass/fail, trust level, primary variable, frozen variables, responsible lead, rollback trigger, CPA, ROAS, Conv. value / cost, post-refund contribution profit, new customer share, refund drift. Proves whether this week can safely hand value to a small tROAS or Maximize conversion value test; still cannot guarantee automation will grow. Release a small test only when trust level is testable and one primary variable changes at a time; do not change tROAS, budget, structure, product groups, and attribution definition together before the value gate passes.

Sample at least 20 recent orders on one row across Ads, GA4, Shopify, and finance

The most dangerous state is not empty value. It is value that looks mostly normal while the business definition has drifted. Your sample table should include order ID, Ads value, GA4 purchase revenue, Shopify gross / net sales, discount, tax, shipping, refund, chargeback, currency, and product group.

Field How to compare Risk signal
Google Ads value Compare row by row with GA4 purchase, Shopify gross / net sales, and finance net revenue Ads revenue rises while backend net sales do not
currency Check USD / CAD / EUR codes and converted values by market Cross-market orders mix currencies and distort ROAS
discount / shipping / tax Confirm discounts are deducted and tax/shipping inclusion follows fixed rules Value looks normal but the business definition is too optimistic
refund / chargeback Connect refunds, chargebacks, high-risk SKUs, and net revenue review High-refund orders keep training the system on gross revenue

Value fault router: locate the fault before scaling it

Value quality problems rarely appear in one dashboard only. Route the symptom first, then decide whether to fix the tag, fix currency, fix deduplication, add refund review, or pause automation trust.

Symptom Likely fault First evidence Do not do
Ads value above Shopify net sales Pre-discount value, tax/shipping-heavy gross value, or refunded orders are still training at original value Sample the latest 20 orders and compare Ads, GA4, Shopify, refunds, and discounts row by row Do not raise tROAS trust while net revenue does not reconcile
Currency drifts across markets Currency is missing, the code is wrong, conversion logic drifts, or account currency and order currency are blended Sample USD / CAD / EUR orders by market and check value, currency, and converted value Do not use mixed-currency blended ROAS to decide cross-market scaling
Duplicate purchase for one order Thank-you page refreshes, multiple tags, duplicate client/server reporting, or unstable transaction ID Pick 10 real orders and compare transaction ID across Ads, GA4, Shopify, and GTM / tag firing records Do not treat duplicate purchase events as conversion-rate improvement
Refund and margin missing from value review The system learns pre-refund gross value and does not know refund rate, margin, chargebacks, or support cost Sample by SKU or product group and connect Ads value, refunds, chargebacks, margin tier, and net revenue Do not let high-refund SKUs keep training automation at full value
Consent and match fields unclear Fields appear only in some payment paths, formatting is inconsistent, or Consent Mode / cookie banner behavior does not match tag behavior Use real checkout paths to inspect field source, pre-hash formatting, consent state, privacy policy, and tag behavior Do not bypass user choice or upload dirty fields for match rate

20oz value trust lab: choose the order issue, then choose the repair action

In a real account, value quality is not one switch. It is a set of order-level evidence. The same 20oz tumbler can expose discount-value drift, Canada currency drift, duplicate purchase after a thank-you-page refresh, and high-refund SKU risk. Do not let better-looking ROAS in Google Ads become the only reason to trust tROAS.

Order scenario First evidence Repair first Write back to the sheet
20oz tumbler discount order Shopify collected $72.80, Ads value received 89.00, and GA4 purchase revenue is 72.80 Fix value definition across discounts, tax, shipping, and pre-refund / post-refund revenue Value gate failed; do not raise tROAS trust
Canada order currency drift Shopify shows CAD 96, Ads value is 96, but currency is missing and the account reads it as USD Fix currency passing and conversion rules Currency gate failed; pause cross-market scale decisions
Thank-you page duplicate purchase The same transaction ID appears twice in GTM preview and GA4 DebugView, while Shopify has one order Fix transaction ID dedupe, GTM triggers, and duplicate client/server reporting Dedupe failed; current value cannot enter automation trust
High-refund SKU keeps spending The 20oz gift set has stable ROAS, but 14-day refund rate is 28% and most tickets mention leaking lids Add post-refund revenue, chargebacks, margin tier, and SKU quality review Post-refund revenue gate failed; fix product quality or review definition first

During the drill, do not default to raise tROAS trust. Only test value-based automation after field source, dedupe, value, currency, post-refund review, and consent boundaries stay stable. Otherwise you are not scaling good orders; you are scaling a bad value definition.

Green diagnostics do not prove the business value definition is correct

Diagnostics can reveal matching, coverage, and implementation issues. They do not tell you whether value should include tax, subtract refunds, or weight new customers. Use diagnostics as a severity signal, not as proof that the business definition is correct.

Severity Signal Action
Low Diagnostics look normal, but business value definition is still under review Keep sampling and do not raise automation trust yet
Medium Match coverage or implementation status fluctuates, while value mostly reconciles Pause optimistic assumptions and inspect fields, tags, and paths
High Order value, currency, dedupe, or refund logic clearly drifts Do not trust tROAS, Maximize conversion value, or weighted value until fixed

Value automation needs trust levels

Trust level Condition Allowed action
Do not trust Purchase, value, currency, transaction ID, or consent boundary is unclear Fix tracking only; no value automation
Low trust Events fire, but sample is low and value definition is still being reconciled Observe conservatively; do not raise tROAS trust
Observation trust Sample orders reconcile and diagnostics show no High risk Keep weekly net-revenue / refund review
Testable trust Fields, dedupe, value, currency, refund, and sampling remain stable Carefully test tROAS or value-based automation with rollback lines

30-minute value QA review: put tech, operations, and finance on one sheet

This lesson should not end with knowing what enhanced conversions are. The usable output is a value QA review that the ad lead, developer, operator, and finance lead can all read. The meeting does not debate whether to trust Google Ads. It checks whether the amount, currency, order identity, and refund treatment being learned by the system match the business facts.

Time Question to confirm Evidence to leave
0-5 min Date window, market, campaign, conversion action, and order sample scope 20-order sample sheet with Ads / GA4 / Shopify / finance sources labeled
5-12 min Whether transaction ID is stable and one order appears once across thank-you page, GTM, and server-side paths Transaction ID export records, order IDs, and trigger-path records for 10 real orders
12-20 min Whether purchase value and currency reconcile row by row across discounts, tax, shipping, and pre-refund / post-refund revenue Value difference notes that separate valid definition differences from implementation errors
20-26 min Whether high-refund SKUs, chargebacks, or low-margin products still train the system at full value Post-refund revenue or SKU quality layer, plus any exclusion, segmentation, or value-weighting decision
26-30 min Whether this week's tROAS state is stop, low trust, observation trust, or testable trust Responsible lead, repair action, review time, and rollback line

If the meeting ends with the data looks okay, the lesson is not finished. A useful closeout is one sentence: because these orders prove value, currency, dedupe, and post-refund revenue are stable, this action is allowed this week; or because this gate failed, the team fixes tracking only and does not hand value to tROAS yet.

The common beginner mistake is treating higher conversion value in Google Ads as business growth, treating enhanced-conversion recording status as a profit signal, or treating one or two days of ROAS movement as permission to scale. Start with three questions: is this the same order value, is this the correct currency, and is this order still worth learning from after refunds?

Copyable lesson notes: turn value QA into next-week accountability

This lesson is easy to misread as a technical switch. Do not only write "enhanced conversions are on." Copy a value trust record instead: whether the system can learn from value, which gate is blocking trust, who fixes it, and when the sample is checked again.

Current pressure: Do not raise tROAS trust just because enhanced conversions are recorded, Diagnostics looks green, or ROAS improved.

First proof: Sample the latest 20 orders and put Google Ads value, GA4 purchase revenue, Shopify gross / net sales, discounts, tax, shipping, refunds, currency, and transaction ID on one row.

Current gate: If match fields, dedupe, value definition, currency, post-refund revenue, Consent Mode, or checkout field source fails, fix signals only and do not scale.

This week's action: Choose one repair move: fix value definition, fix currency, fix transaction ID dedupe, add post-refund revenue, or keep the account in low-trust observation.

Stop action: Do not let Maximize conversion value or tROAS scale this value set while net revenue, currency, duplicate purchase, post-refund revenue, or consent boundary fails reconciliation.

Admin fields: Write Enhanced conversions / Diagnostics, Google tag / GTM, Consent Mode, Ads value, GA4 purchase, Shopify Orders / Refunds, finance net revenue sheet, SKU / GTIN / product group, Bid strategy report, and Change history into one value QA record; include at least conversion action name, primary / secondary, enhanced conversions status, ad_storage, ad_user_data, transaction_id, Ads value, GA4 revenue, Shopify net sales, currency, refund amount, post-refund revenue, contribution profit, 20-order sample pass/fail, trust level, and rollback trigger.

Review window: Re-sample on the same 7-day basis next week. Prove the result with order IDs, field values, diagnostics status, and change records, not vague dashboard impressions.

Next route: If the value gate passes, move to when and how to scale. If it fails, return to conversion tracking setup and metric reading to complete the evidence.

Stop / Go rules: do not let tROAS scale wrong value

Stop Go Proof needed
Order value cannot reconcile to Ads value Sample orders explain Ads / GA4 / Shopify / finance differences Last 20 orders reconciliation sheet
Currency drifts, purchase duplicates, or transaction ID is unstable Currency and transaction ID are stable in real order paths Cross-market order samples and tag test records
High-refund orders continue training the system at full gross value Refund, chargeback, and net-revenue review are part of the value definition Post-refund revenue or finance review layer
Diagnostics and business checks repeatedly conflict Diagnostic severity, responsible person, fix action, and review window are written Value-signal stop/go record
Back to Course Outline
11
View All Tutorials

After copyable notes

Connect this lesson to the next learning and membership path

Copyable notes are not a download pack. Their job is to carry the decision, evidence, and next action out of the lesson. Continue to the next lesson first; if this page solved a real problem, check whether the member tutorial path can close the rest of the workflow.

Share this tutorial

If this lesson helped, send it to a teammate, friend, or partner before moving on to the next one.