Table of content
You chose Odoo to run your sales, your quotes and your customers. Your website, on the other hand, may run on Shopify, WooCommerce, PrestaShop, Magento, Drupal, Wix, a Node.js application, a custom PHP build, or on Odoo itself. And your ad platforms, Google Ads, Meta, TikTok, LinkedIn or Microsoft Ads, only see what the browser is willing to pass on. The result: three different figures for the same sale, B2B leads whose origin nobody can trace any more, and campaigns that learn from partial data.
This guide describes the method A-Track applies to B2C and B2B companies that use Odoo as their CRM or ERP: collect cleanly on every site, consolidate in Odoo, then send a single, deduplicated and compliant dataset back to the ad platforms and dashboards. It is written for business owners and marketing leads, e-commerce managers, and the web agencies and Odoo integrators who build these ecosystems.
Key takeaways
Odoo is the best source of truth for your marketing: it knows real sales, won quotes and B2B pipeline stages, which no ad platform can see.
Out of the box, Odoo keeps a single source per record: the last UTM link clicked within 31 days, with no click ID and no history. A-Track's attribution scripts keep the first source, the second, the last and the dated history, as Odoo tags.
Whatever the CMS, the principle stays the same: a clean data layer, web GTM with a CMP and Consent Mode v2, then server-side GTM hosted on Stape, first-party.
From Odoo, the Stape app for Odoo CRM turns pipeline stages into conversions and feeds Customer Match and Meta audiences.
Four deduplication locks stop a sale from being counted twice: browser and server, contacts, orders, ad platforms.
In Switzerland, Google does not match on IP address and timestamp: the quality of the emails and phone numbers collected with consent determines the match rate.
Why should Odoo become the source of truth for your marketing?
An ad platform only knows what it measures itself: a click, then a conversion reported by a tag in the browser. Everything that happens elsewhere escapes it. Ad blockers and browser protections cut part of the signals. Visitors who refuse advertising cookies drop out of named tracking. Payments and bookings on an external domain break the chain, as we explain in our article on cross-domain tracking. Above all, much of the value is created off the site: a B2B order confirmed by phone, a quote signed three weeks after the request, a customer who comes back to the shop.
Odoo sees all of this. It records opportunities and their stages, quotes, orders, invoices, cancellations and returns, and the loyalty of a customer over several years. It is the only database that knows how much you really sold, to whom, and at what margin.
The consequence is simple: instead of adding up each platform's figures, you need to set them against Odoo's revenue. Google Ads and Meta each claim their share; Odoo knows the total. In B2B the stakes are even higher. Without feedback from the CRM, the algorithms optimise for form submissions, not for signed customers. The whole point of offline conversions is to send them what happened after the click.
What does the A-Track architecture around Odoo look like?
The architecture has four layers, always the same, whatever the technology behind the site.
Sources. One or more e-commerce sites, lead generation sites, and sometimes marketplaces that report their sales via API.
Collection. Google Tag Manager in the browser, driven by a consent management platform (CMP) and Consent Mode v2. A-Track's attribution scripts capture where each visit comes from.
Server and CRM. Server-side GTM, hosted by Stape on a subdomain of your site, receives the events, enriches them and filters them. Odoo receives contacts, requests and orders together with their sources.
Activation. Ad platforms receive consolidated conversions, audiences are synced via API, and reporting finally compares what each platform claims with real revenue.
Two flows connect the server and Odoo. In one direction, the journey history reaches the contact record. In the other, Odoo sends the server the conversions only it knows about: a won quote, a paid order, a pipeline stage reached.
How do you capture data on your site, whatever the platform?
The tracking plan does not change from one site to another. The same events, named the same way, describe the e-commerce journey (product view, add to cart, checkout start, purchase) and the lead generation journey (contact request, quote request, sign-up, download). What changes is how the data layer, the layer of data the site exposes to GTM, gets populated. We explain why a data layer beats reading the page: it is what lets tracking survive a redesign.
Stape publishes apps for most of the CMSs on the market. They follow the same pattern: they feed the data layer with e-commerce events in GA4 format and, in parallel, send order webhooks from the site's server. These webhooks catch purchases where the customer never returns to the confirmation page, but they carry no visitor cookies: they complement the browser event, they do not replace it. On the Odoo side, remember one point: there is no official Odoo connector for Shopify, WooCommerce, PrestaShop, Magento or Wix. Orders arrive through a third-party app or the API.
WooCommerce
The free Stape Conversion Tracking plugin feeds the data layer (product view, cart, checkout start, purchase, sign-up, login), adds user data and sends webhooks for orders created, paid or refunded. Two pitfalls come up often. Since WooCommerce 8.3, block-based cart and checkout are the default, and a payment plugin that has not been adapted can break tracking. And some gateways never bring the customer back to the thank-you page: that is where the webhooks take over. WooCommerce also records its own order attribution, useful for cross-checking figures.
Shopify
The Stape app for Shopify installs GTM on the store and adds a custom pixel in Customer Events for the checkout. The checkout is sandboxed: scripts cannot see the page, events do not show up in GTM preview, and consent given on the store does not always carry through to checkout. The CMP choice therefore has to be relayed, through Shopify's Customer Privacy API or a consent update tag. The old additional scripts on the thank-you page can no longer be edited: everything now goes through pixels. Our comparison of Shopify and WooCommerce server-side details these differences.
PrestaShop
The Stape module for PrestaShop pushes e-commerce events and sends an order webhook from the back office, for payments that never return to the confirmation page. As on other CMSs, this webhook carries no cookies: the browser event remains essential to tie the sale to its campaign.
Magento and Adobe Commerce
The Stape extension for Magento 2 pushes events, user data and webhooks triggered by the order status of your choice. The tricky part is the full-page cache: visitor-specific data, such as the cart or the logged-in customer, is loaded separately in the browser. The data layer must read it there, not in the cached HTML, or it will send another visitor's information. The native GTM integration exists only in Adobe Commerce, not in Magento Open Source.
Drupal
The widely used google_tag module loads GTM but does not define e-commerce events. Stape has just released a Drupal module, still very little used. On Drupal Commerce, as on a Drupal lead generation site, the data layer is therefore usually custom-built. On the Odoo side, no connector is listed: you go through the API or webhooks.
Wix
Wix strictly limits custom code: GTM and ad pixels do not work as manually added code. The native GTM integration requires a Premium site with a connected domain, and pushes its events in the old Enhanced Ecommerce format, which has to be converted for GA4. The Stape app for Wix injects GTM, pushes events in GA4 format and sends order webhooks.
Custom Node.js and PHP
On a custom-built site, the data layer is written with the development team, based on the tracking plan. For events that originate on the server, such as an order confirmed after an asynchronous payment or a renewed subscription, Stape publishes PHP and Node.js SDKs that send events straight to the server-side container. On the Odoo side, the external API makes the link. Odoo 19 introduced a new API and has announced the end of the old XML-RPC and JSON-RPC APIs: any new development should take this into account.
Odoo Website and eCommerce
When the site runs on Odoo, Odoo's own documentation says it plainly: data layers are not managed there. Stape's free app for Odoo eCommerce fills that gap. It pushes the main GA4 events, from product list to purchase, adds hashed user data and sends purchase and refund webhooks from Odoo, useful when a payment gateway does not bring the customer back to the site. The site's forms, for their part, can create opportunities directly in the CRM. Note: like every module from the Odoo App Store, this app requires Odoo.sh or an installation on your own servers, because Odoo Online does not accept these modules.
And marketplaces such as Amazon?
No tag runs on Amazon's pages. Sales come in through Odoo's official Amazon connector, which syncs orders whether they are fulfilled by Amazon or by you. But Amazon's API does not pass on the buyer's email, phone number or name. These orders therefore enrich consolidated revenue and product analysis, without joining named audiences or the journey history.
How do you keep each contact's original source in Odoo?
This is the most underestimated point, and the one that changes the reading of your marketing the most.
What Odoo does out of the box
Odoo has a UTM module. Opportunities, quotes, orders and invoices carry three fields: campaign, source and medium. When the site runs on Odoo, the utm_campaign, utm_source and utm_medium URL parameters are copied into cookies that live for 31 days, then copied into the record created from the site. This behaviour is not described in the documentation; it comes from our reading of the Odoo 18 and 19 source code.
Three limits follow:
One value per field. Each new UTM link overwrites the previous one. The record keeps the last link followed within 31 days, and the source may even come from a different visit than the campaign.
No click ID. Odoo does not natively capture Google Ads' gclid or Meta's fbclid, without which an offline conversion is poorly matched to its click.
No history. The first source, often the most useful one to judge an acquisition campaign, is lost as soon as a second one arrives.
These cookies are classed as optional: Odoo's cookie banner blocks them until the visitor accepts. And when the site is not on Odoo, they simply do not exist.
What A-Track's attribution scripts add
When a visitor lands on the site, A-Track's scripts read everything that describes where they come from: UTM parameters, referrer, landing page, date, and the ad platforms' click IDs (gclid, gbraid and wbraid for Google, fbclid for Meta, ttclid for TikTok, msclkid for Microsoft, li_fat_id for LinkedIn). This information is kept first-party, in line with the choice made in the CMP.
When the visitor becomes a contact, by sending a form, asking for a quote or placing an order, this data is passed to Odoo through hidden fields or order attributes. There it becomes fields and tags:
Source 1: the contact's first known origin;
Source 2: the second, often the remarketing or email that revived interest;
Last source: the one before the purchase or request, excluding direct visits;
History: every touchpoint, dated, visible on the record.
Why tags? Because they can be filtered, combined and reused everywhere in Odoo: pipeline views, sales reports, email segments, and even predictive lead scoring, which can take source and tags into account. In B2B, where a sales cycle lasts several months, this is often the only way to know which campaign brought in a customer who signed six months later.
How do you avoid counting the same sale twice?
The more sources an architecture has, the greater the risk of duplicates. A-Track puts four locks in place.
First lock: browser and server
When the same conversion is sent by the browser tag and by the server's conversions API, both carry the same event ID. The platform keeps only one. Our article on server-side Meta CAPI details this mechanism.
Second lock: contacts
The same customer can exist three times in Odoo: once with an upper-case email, once with a badly formatted phone number, once as the B2B contact of their company. A-Track normalises identifiers before sending them (email in lower case without spaces, phone in international format) and relies on Odoo's own tools: the similar leads button in the CRM, which matches records by email and phone, and Odoo Enterprise's Data Cleaning app, whose deduplication rules run every day. Be careful: when records are merged, Odoo keeps only one value per UTM field. One more reason to keep the source history in tags, which add up.
Third lock: orders
An order can reach Odoo through an e-commerce connector and, in parallel, through a purchase event sent to the tracking server. Each order therefore keeps its original reference, specific to its channel, and each destination has a single source of truth for a given purchase: either the site event or Odoo's confirmation, never both.
Fourth lock: ad platforms against each other
Google Ads and Meta can both claim the same sale, each with its own attribution window. Their total then exceeds real revenue. This is not a bug, it is how they work. The only answer is reporting that starts from Odoo: real revenue as the reference, and next to it, what each platform claims.
How do you send reliable conversions to ad platforms from Odoo?
Stape publishes an app for Odoo CRM, Conversion Tracking for CRM, available for Odoo 19 on Odoo.sh or on your own servers (Odoo Online does not accept App Store modules). As soon as an opportunity reaches a stage you have chosen, the app sends the matching event. You map each pipeline stage to an event name, or send every stage and exclude the ones you do not need. Contact data is taken from Odoo fields, and the consent status is sent along with it.
According to Stape's documentation, the app offers five sending methods that can be combined: to your server-side GTM container, to Google Ads Customer Match, to Meta custom audiences, to Meta's Conversions API Gateway, and to Stape's Signals Gateway. For A-Track, the reference route goes through server-side GTM: that is where the same consent, hashing and deduplication rules apply to every platform.
From the server, each platform receives conversions through its own channel:
Google Ads: enhanced conversions and offline conversions. Google has announced that from 15 June 2026, offline conversion import and enhanced conversions for leads move to the Data Manager API; we cover this change in our article on Google Ads Data Manager.
Meta: the Conversions API, including CRM events that mirror your lead stages. The optimisation goal dedicated to quality leads does, however, require Meta instant forms and enough volume.
TikTok: the Events API, server to server.
LinkedIn: the Conversions API, valuable in B2B for reporting qualified leads.
Microsoft Ads: a Conversions API exists, still being opened account by account at the time of writing.
Two rules make the difference. First, send an accurate value: the real order amount, or a value per B2B pipeline stage, rather than a fixed value per form. Second, send the click ID captured by A-Track's scripts in addition to the hashed email and phone number: it is what most reliably ties a sale to a campaign. As for ads in ChatGPT, OpenAI announced a pixel and a conversions API in 2026, but the opening of this platform to the Swiss market remains to be confirmed.
Which audiences should you build from Odoo?
Odoo knows your best customers, your recent customers, your open prospects and your B2B accounts. Synced via API, these lists become far more precise audiences than those built from site visitors alone.
Local VIP: the best customers in a region, for local campaigns or lookalikes.
Remarketing: prospects who asked for a quote without signing, customers whose last purchase was months ago.
Exclusion: customers who have just bought, orders in progress, accounts already signed, so you stop paying for useless clicks.
Cross-channel: the same segment pushed to Google, Meta and TikTok, for a consistent message.
Four rules govern these audiences, set out in particular in Google's Customer Match policy. Membership in a Customer Match list lasts 540 days at most. Emails and phone numbers must be normalised then hashed with SHA-256 before sending, as Google and Meta require. In the EEA, the UK and Switzerland, Google does not support matching on IP address and timestamp: the quality of the contact details collected determines the match rate. Finally, only contacts who have given their consent should join an audience, and your privacy policy must say so.
How do you measure Odoo email marketing as a real channel?
For many companies on Odoo, email is one of the most profitable channels, and one of the worst measured. Yet Odoo's Email Marketing app tracks a lot: sends, deliveries, opens, clicks, replies, bounces and unsubscribes. It automatically adds UTM parameters to links, with "email" as the default medium. Each mailing shows the number of leads, quotes and the invoiced amount attributed to it, and an A/B test can pick its winner on these criteria.
These figures do, however, share the limits of the UTM module. Attribution relies on the source stored in the record, so on the last touch within 31 days. A customer won through Google Ads then nudged by a newsletter will be credited to the newsletter, and the acquisition campaign will never know. Open rates are also inflated by email clients that preload images.
A-Track aligns the UTM naming of mailings with that of paid campaigns, adds every email click to the source history, and builds reporting that separates the revenue generated by email as first, second and last touch. Email is then judged as a real channel, alongside the ad platforms.
What do the FADP and the GDPR say about this setup?
A server-side setup does not exempt you from any rule; it lets you apply them the same way everywhere.
Consent first. The CMP drives Consent Mode v2 and its two advertising signals, ad_user_data and ad_personalization. Google's EU user consent policy explicitly covers Switzerland. Our Consent Mode v2 guide details the setup.
A single banner. If the site runs on Odoo, Odoo's banner and the CMP must be coordinated, so that Odoo's UTM cookies follow the same choice as everything else.
Hashing is not anonymisation. A hashed email remains personal data under both the FADP and the GDPR: sending it needs a clear basis and must appear in your privacy policy.
Control of data flows. Server hosting, contracts with processors, record of processing activities: our guide to server-side tracking and the FADP covers these points.
Who does what between the Odoo integrator, the web agency and A-Track?
These projects succeed when everyone stays in their role.
The Odoo integrator handles connectors, the data model, the custom fields that receive the sources, and the deduplication rules.
The web agency exposes the data layer, adds hidden fields to forms and makes sure the checkout passes the right attributes.
A-Track designs the tracking plan, deploys the CMP, web GTM and server-side GTM, installs the attribution scripts, configures the Stape app for Odoo, then tests and monitors the whole setup.
The media agency chooses campaign goals, conversion values and how audiences are used.
Server-side tracking degrades as soon as a plugin, a theme or a pipeline stage changes. We explain how to monitor it after launch.
Where should you start?
An audit of the current setup: tags, consent, flows into Odoo, gaps between the platforms and real revenue. Our tracking and compliance audit starts with a flash audit.
A tracking plan shared by all sites, with the UTM naming convention and the list of Odoo tags.
Deployment of the CMP, web GTM, server-side GTM and the attribution scripts.
Connecting Odoo: fields, tags, Stape app, audiences.
Testing and maintenance, so the figures stay accurate over time.
You use Odoo and want data your campaigns can truly rely on? Discover our Tracking and Compliance service.