One business event, many platforms
The tags, triggers and variables guide recommends the pattern in
one sentence: the same custom event trigger for purchase serves the GA4 event tag and the ad platform
conversion tags alike. The reason is consistency. When every tag reads the same push, every platform sees the
same order at the same moment with the same value, and a difference between platforms is a difference in
their attribution, not in what was sent.
The alternative is the common case. A Google Ads tag fires on the thank-you page view, the Meta pixelAn ad platform's browser code that reports page views and events on a site back to the platform. reads the order total from the page's text, the TikTok tag fires on a button click, and GA4 reads the data layer. Four definitions of one purchase produce four counts, and nobody can say which differences are real. The data layer guide describes the remedy at the level of one event; this tutorial applies it across platforms. The academy exercise on mapping one data layer to every platform for a cookery school practises writing that event contract and mapping it to analytics and three ad platforms with a shared event ID for deduplication.
Before you start
- The data layer in place, with the contract and timing rules of the data layer guide, and a developer who can add an identifier to the purchase and lead pushes.
- The building blocks: custom event triggers, data layer variables, constants and lookup tables, from the tags, triggers and variables guide.
- Each platform's account set up in its own section: the Meta dataset, the TikTok pixel, the Google Ads conversion actions created as the Google Ads conversion tracking guide describes, and any other platform's conversion definitions. This tutorial wires them; it does not create them.
- A consent setup that sets defaults before any tag, as the consent mode guide describes.
- An inventory of the existing platform tags, so the old ones can be removed in the same release the new ones go live.
- Access to each platform's testing tools and a way to place a test order and a test lead.
List the conversions each platform needs
Start from the business, not the platforms. Write down the actions that matter (a purchase, a lead, a sign-up, perhaps a basket addition for shops) and, for each platform, which of them it optimises towards or reports. Google's page on conversion tracking describes conversions as the actions you count as valuable to the business, and the same discipline applies everywhere: a platform that optimises towards an action should get the real outcome, not a proxy chosen because it fires more often. Keep the list short. Every extra action is a mapping row, a tag per platform and a test.
Write one data layer event per business action
For each action, agree one push with the developers, sent when the action is true: after the order is confirmed, after the server accepts the lead. Include what every platform will need: the value as a number, the currency code, the order or lead identifier, a separate event identifier for deduplication, and the items for commerce events. Keep customer data out of this shared push. Where a platform feature needs hashed customer data, such as Enhanced ConversionsA Google Ads feature that sends hashed first-party customer data to improve conversion matching. in Google Ads or advanced matching on TikTok, scope those fields to the tags that need them, as the guide to keeping personal data out of analytics explains, so the analytics tag never reads them.
window.dataLayer = window.dataLayer || []; window.dataLayer.push({ ecommerce: null }); window.dataLayer.push({ event: "purchase", event_id: "ord-58213", ecommerce: { transaction_id: "58213", value: 64.00, currency: "GBP", items: [{ item_id: "SR-114", item_name: "Night serum", price: 32.00, quantity: 2 }] } });Build the mapping table
Make a table with one row per business event and one column per destination. In each cell, write the event the platform documents for that action and the fields it takes, using only names stated on the platform's own help pages or on this site's platform guides: GA4's recommended events from the events and parameters guide, Meta's standard events from the pixel tutorial, TikTok's events from its tutorial, and the Google Ads conversion action and its label. Note where each name came from. Store the table in the tracking plan, next to the data layer contract, because it is the specification the container is built and audited against. The illustrative table after the steps shows the shape.
Create shared variables
Build each value once and let every tag read it. Data layer variables for value, currency, transaction identifier, event identifier and items. Constants for every account identifier: the Meta dataset, the TikTok pixel code, the Google Ads conversion ID. Lookup tables where one input maps to a platform-specific output, such as a business event name to a Google Ads conversion label, which the building blocks guide gives as a standard use. And one custom event trigger per business event, shared by every tag that reports it. Name everything by the container's convention so a reviewer can see, in the variables list, that the Meta tag and the GA4 tag read the same value.
Build each platform's tag from the mapping
For each cell of the mapping table, build one tag: the platform's event name from the table, value and currency from the shared variables, the content or item fields the platform documents, and the event identifier in the field the platform provides for it. Prefer the platform's template from the community gallery to custom HTML. Make sure each platform's base code loads before its event tags on the same page, by tag sequencing or the setup the platform's tutorial describes, and put each platform's tags in a folder of its own. Then remove the old hand-built tags for the same actions in this workspace, so the release swaps one set for the other rather than adding to it. For LinkedIn, the tutorial on setting up the Insight Tag and conversions covers the base tag and the URL or event-specific conversions that the mapped events feed.
Apply consent checks to every tag
Google's tags carry built-in consent modeA Google tag feature that adjusts what tags collect according to the visitor's consent choices. checks; every other tag needs its consent settings filled in. The consent mode guide explains that a non-Google tag can require additional consent types, such as
ad_storage, before it fires, and that the consent overview shows every tag's status in one view. Set the requirement on every ad platform tag, base and event alike, to match the categories your consent platform uses. The Meta and TikTok tutorials describe the alternative pattern of firing on a trigger the consent platform raises after a positive choice; choose one pattern per container, not one per tag, and check the consent overview before publishing.Share an event ID with server-side sends
Where a platform receives the same action from the browser and from a server, it needs a way to keep one copy. Meta documents event name and event identifier as the primary deduplication keys for the pixel and the Conversions APIServer-side event sharing with Meta., as the Conversions API guide explains, and TikTok documents the same pairing for its pixel and Events API. Pass the
event_idfrom the data layer into each browser tag's event identifier field, and have the server read the same value from the order or lead record for its payload, with the same event name. Send the server copy promptly, because both platforms deduplicate within a window. The pixel versus Conversions API comparison explains why both routes run together, and the duplicate conversions diagnosis shows what a mismatch looks like.Test each platform's receipt
In Tag Manager preview, place a test order and a test lead and confirm that each business event fires every mapped tag once, with the same value, currency and identifiers, and that no old tag fires alongside. Then check receipt in each platform's own tools, as its guide on this site describes: the test events view and Pixel Helper for Meta, the test events view, pixel helper and test event code for TikTok, the conversion checks in the Google Ads section's tracking guide, and the tag and event checks the LinkedIn Insight Tag diagnosis works through. Where server sends are live, confirm that the browser and server copies are marked as deduplicated. Repeat with consent refused: nothing from the ad platforms should leave the browser.
Publish and watch the first week
Publish with a version note naming the tags added and removed and the tests passed. On each day of that first week, compare each platform's count of each action with the order or lead system for the previous closed day, and read Meta's and TikTok's deduplication status. Expect each platform to sit near the real count rather than on it: consent and blockers remove some events, and each platform's reports count by its own attribution rules. A platform near double the real count points at failed deduplication or an old tag still firing; one far below points at a trigger or consent problem. Fix in a new workspace, and record each finding in the tracking plan.
A mapping table (illustrative)
Cindermoss Skincare's purchase and sign-up mapping
Cindermoss Skincare, an invented skincare brand, had four sets of hand-built tags: a GA4 purchase tag on the data layer, a Google Ads tag on the thank-you page view, a Meta pixel reading the order total from the page, and a TikTok tag added by an agency on the "place order" button. Meta and TikTok also received purchases from the shop platform's server integration, with identifiers generated separately on each side. The rebuild mapped two business events to every destination.
| Business event | GA4 | Google Ads | Meta | TikTok |
|---|---|---|---|---|
purchase | purchase with value, currency, transaction ID, items | "Purchase" conversion action, label from a lookup table, value, currency, transaction ID | Purchase with value, currency, contents, event ID | CompletePayment with value, currency, contents, event ID |
sign_up | sign_up with method | "Account sign-up" conversion action | CompleteRegistration with event ID | CompleteRegistration with event ID |
The developers added event_id to both pushes, set to the order number with a prefix for purchases and the
account record's identifier for sign-ups, and the shop platform's server integration was reconfigured to send
the same values. Across a closed fortnight with 2,140 orders in the shop:
| Destination | Purchases reported before | Per real order | Purchases reported after | Per real order |
|---|---|---|---|---|
| Meta | 3,660 | 1.71 | 2,230 | 1.04 |
| TikTok | 3,320 | 1.55 | 2,180 | 1.02 |
| Google Ads | 2,590 | 1.21 | 2,010 | 0.94 |
| GA4 | 1,960 | 0.92 | 1,970 | 0.92 |
Meta and TikTok had been counting most orders twice, once from each route. Google Ads had been counting reloads of the thank-you page. GA4, already on the data layer, barely moved. The remaining gaps above and below one order are each platform's attribution of what it could match, which the team now reads as a real difference rather than a tagging one.
Keeping platforms in step
The mapping table only helps if it stays true. Three habits keep it that way.
Change the contract first. A new business event, a new field or a change in when a push fires starts with the data layer contract and the mapping table, then the tags. A change made in one platform's tag alone is how the four definitions return.
Add and remove platforms as columns. A new platform is a new column in the table, a new folder in the container and a receipt test; the site does not change. A platform the business stops using is removed the same way, its tags paused and then deleted as the container governance guide describes for anything no longer used, rather than left in the container indefinitely.
Audit the container against the table. At each container review, compare every ad platform tag with its cell in the mapping table: the same trigger, the same variables, the event identifier present. Any tag with no cell is either a gap in the table or a tag to remove. The Tag Manager hub links the governance and debugging guides that make the review routine.
Platform settings stay where they belong. Attribution windows, optimisation events and conversion action settings are configured in each platform and covered in its own section of the site; the mapping table makes sure each platform receives the same facts, and the platform decides what to do with them. The academy checkpoint on data layer contracts, embedded tools and User-ID tests why one business event mapped to many platforms beats each platform's own trigger, and how shared event IDs deduplicate browser and server events.
Common mistakes
- One data layer event per platform. Developers maintain four pushes for one purchase, and they drift.
- Values read from the page. A pixel scraping the order total breaks at the next redesign and reads the basket total with a currency symbol.
- A random event ID generated in the browser. The server cannot know it, so deduplication never matches.
- New tags added, old tags left. The release that was meant to replace the hand-built tags doubles every conversion instead.
- Customer data in the shared push. Hashed or not, it ends up readable by every tag, analytics included.
- Consent set on some tags only. One platform keeps firing for visitors who declined advertising.
- Platform settings fixed in the container. Attribution and optimisation belong in each platform; the container's job is to send the same facts to all of them.
- No receipt test per platform. A tag that fires in preview can still be rejected by the platform for a wrong identifier or a missing currency.
