Connecting CRM data and customer value to Google Ads: a step-by-step setup
Updated: 29 July 2026 Short answer: without this, the algorithm optimises for form fills rather than sales If you sell on a lead model and the only signal reaching Google Ads is “someone submitted a form”, the bidding system does exactly what you asked: it finds people inclined to submit forms. It has no signal […]
Updated: 29 July 2026
Short answer: without this, the algorithm optimises for form fills rather than sales
If you sell on a lead model and the only signal reaching Google Ads is “someone submitted a form”, the bidding system does exactly what you asked: it finds people inclined to submit forms. It has no signal that would let it distinguish a lead that becomes a contract from a lead that never answers the phone.
Offline conversion imports close that gap. They let Google Ads connect what happens after the click, meaning answered phone calls, closed deals and signed contracts, to the original keyword and campaign that drove the lead. Without this data, Smart Bidding is flying blind, optimising for cheap form fills instead of profitable customers.
There is a more urgent reason to address it now. The technical path changed in 2026 and some existing integrations have stopped working. The UploadClickConversions API request used by custom integrations was deprecated on 15 June 2026 and must move to the Data Manager API. If your integration was built before 2024 and nobody has touched it since, audit which path you are on this week.
What offline conversion import does
The mechanism runs on a click identifier. Google writes it when a user arrives from an ad, you store it in your CRM alongside the lead, and when the lead reaches a valuable status you send that information back to Google Ads.
The effect is a change in what the bidding system learns. Instead of optimising lead count, it starts optimising lead quality, because it now receives a signal about which sources generate transactions.
Google offers its own figure in favour of the enriched variant: advertisers who used first-party data such as email addresses and phone numbers alongside GCLIDs imported through offline import saw a median 10% increase in conversions compared with those using standard offline conversion imports. Note that this is platform-vendor data rather than independent research, so read it as a direction rather than a promise.
Which path is correct today
This question has a different answer in 2026 than it did two years ago. Three orienting facts:
- Click-identifier-only import is the legacy path. GCLID-only import is now Google’s legacy path, and Enhanced Conversions for Leads is the recommended setup, reporting more accurately.
- Data Manager is the configuration hub. The fastest and easiest way to set up or upgrade to enhanced conversions for leads is by using Data Manager, where GCLID and user-provided data can both be used as match keys for data imports.
- Custom integrations require migration. From 15 June 2026 the UploadClickConversions request is no longer the path forward for certain offline click conversion import workflows, including workflows tied to enhanced conversions for leads, and Google is steering advertisers and developers toward the Data Manager API and Data Manager-based ingestion options.
Also worth checking whether you rely on a retired integration: the legacy Salesforce integration has not been supported since 31 May 2025.
Delivery methods, most to least automated
Google lists several routes and ranks them clearly: an automated method lets you best utilize Smart Bidding by ensuring your conversion data is up to date. You can send it automatically using Data Manager or through the Google Ads API, or use the integration with Zapier or Salesforce to connect your CRM to Google Ads. If you prefer to upload from files, there is an SFTP connector in Data Manager that doesn’t require data preparation or a fixed template, and for manual uploads a simple five-column Excel or CSV template.
The choice depends on your stack and cadence. The practical rule: the less often you send data, the slower the system learns, and a manual monthly upload is usually worse than an automated daily one.

Step-by-step setup
Step 1: capture the click identifier in your CRM
This is where the whole process most often breaks. The identifier has to be written on every form submission, usually into a hidden field, and it has to survive every redirect on the way to the CRM.
Three typical causes of low match rate: users disabling cookies before clicking, the GCLID being stripped by a redirect, and the CRM not saving GCLIDs for all form submissions. A reference point for a working setup: a match rate of 75 to 85% is typical, and anything below 50% signals a problem.
Step 2: define funnel stages as separate conversion actions
The platform recommendation here is unambiguous: it’s best practice to create a separate conversion action for each stage of the funnel, which gives more clarity in reporting and allows for better segmentation.
In practice that usually means three or four actions: lead, qualified lead, opportunity, closed deal. Each with a different value reflecting probability of closing.
Step 3: prepare the CRM export
The file structure is simple but has two places where uploads reliably fail.
Column set and a sample row: click identifier, conversion name, conversion time, value, currency, producing a row like abc123xyz, Closed Won Deal, 2026-05-10 10:00:00 +0000, 5000, USD. Conversion value is optional but highly recommended for Target ROAS bidding, and should be a numeric value without currency symbols.
The two most common errors:
- Timezone formatting. This is where many uploads fail, because the timezone offset must follow the format +0000 or -0500 with no colon between hours and minutes.
- Records without an identifier or outside the time window. The recommended export filter: filter your CRM for leads that have reached your conversion milestone, such as a status of “Closed Won” or “Qualified Call Completed”, then filter further to include only records where the GCLID field is populated and where the conversion date falls within 90 days of the lead creation date.
A note on the window: published figures differ between sources, and defaults depend on account configuration. Check the conversion window in your own account rather than adopting a number from a guide.
Step 4: assign values, not just events
This is the step that separates a correct configuration from one that actually changes results. An event without a value tells the system “this happened”. An event with a value tells it “this was worth this much”, which allows bidding toward revenue rather than volume.
Values do not have to be actual revenue. With a long sales cycle, expected values work better: average deal value multiplied by the historical probability of closing from that stage.
Step 5: do not switch to a primary goal immediately
Do not mark this conversion as a primary conversion goal yet. Observe for a few weeks first, checking that data arrives and that the match rate sits in a normal range. Switching a bid strategy onto an unverified data stream resets the learning phase and usually degrades performance for several weeks.
How much this actually changes
Published single-implementation figures need careful reading, but they show the mechanism. A documented case from a lead generation agency serving a B2B SaaS client on a $3,000 monthly budget: target CPA campaigns were hitting a $45 cost per lead, but only 12% of those leads converted to paying customers, because Smart Bidding kept optimising for the cheapest leads rather than the most qualified. After implementing GCLID capture in a HubSpot hidden field and weekly uploads of MQL and Closed Deal events at different values, $50 for MQL and $500 for a deal, within six weeks Smart Bidding shifted spend toward keywords generating MQLs, deal volume rose 34%, and cost per deal fell from $380 to $255.
Caveat: this is a single agency-reported case, without a control group, on a small budget. Treat the direction as credible and the magnitude as non-transferable. Measuring a change like this properly requires an incrementality test, not a before-and-after comparison.
Audit checklist for an existing setup
- [ ] Establish which technical path you are on: identifier-only import, Enhanced Conversions for Leads, or a custom API integration
- [ ] If custom: confirm migration to the Data Manager API following the June 2026 deadline
- [ ] Check whether you rely on the retired Salesforce integration
- [ ] Measure your match rate and compare against the 75 to 85% range
- [ ] Verify the click identifier is written on every form submission, not just some
- [ ] Verify that on-site redirects do not strip the identifier
- [ ] Verify a separate conversion action exists for each funnel stage
- [ ] Verify conversions carry a value, not just an event
- [ ] Review upload frequency and consider moving to automated delivery
- [ ] Verify the bid strategy actually uses the imported conversions as its goal
The identifier capture and redirect items sit in the tagging layer, so they usually belong to a Google Analytics and GTM audit rather than to work inside the ads account. Configuring conversion actions, values and bid strategy is part of ongoing Google Ads management.
Frequently asked questions
Is offline conversion import GDPR compliant?
Sending first-party data requires a legal basis and appropriate consent, and contact data is hashed before transmission. Enhanced Conversions for Leads is positioned by the platform as privacy-adapted, but assessing compliance in your specific case belongs to your legal team, not to the tool vendor.
What if my CRM does not store click identifiers?
That is the first thing to fix, and it usually means adding a hidden form field and confirming the value survives every redirect. Without it, no downstream configuration will work.
Can I import conversions if I sell over the phone?
Yes, and this is one of the most common use cases. The requirement is linking the call to the session that created the lead, which usually means capturing the identifier at first contact and storing it in the call handling system.
How long after a click can a conversion be imported?
The window depends on how the conversion action is configured in your account. Published guides give different figures, so check the setting in your own account rather than adopting a number from an article. Records outside the window will not match.
Is this worth doing at low volume?
Yes, but the effect will be slower. Bidding systems need data, and at a dozen or so transactions a month the signal is too sparse to change algorithm behaviour meaningfully in the short term. At that scale the value of the import is mainly in reporting rather than optimisation.
Do conversion values have to be actual revenue?
No. With a long sales cycle, expected values work better: average deal value multiplied by the probability of closing from that stage. What matters is that the proportions between stages reflect real differences in value.
Sources
- Google Ads Help, Offline conversion imports FAQs → https://support.google.com/google-ads/answer/10029210?hl=en
- Google Ads Help, About offline conversion imports → https://support.google.com/google-ads/answer/2998031?hl=en
- ALM Corp, Google Ads Offline Conversion Imports in 2026: June 15 API Change, Data Manager Migration, May 2026 → https://almcorp.com/blog/google-ads-offline-conversion-imports-2026-data-manager-api/
- Farsiight, Google Ads Offline Conversion Tracking in 2026: What Changed and What to Do, June 2026 → https://www.farsiight.com/resources/offline-conversions-google-ads/
- Cometly, Google Ads Offline Conversion Import: Complete Guide, May 2026 → https://www.cometly.com/post/google-ads-offline-conversion-import
- NPPR Team, Google Ads Offline Conversion Imports: Setup Guide 2026, April 2026 → https://npprteam.shop/en/articles/google/google-ads-offline-conversion-imports-setup-2026/
Data currency and limitations note. Migration deadlines and available delivery methods change on a several-month cycle, and some features roll out in stages across accounts. Confirm the current position in Google Ads documentation before using this article as an operating instruction. The cited 10% conversion increase comes from the platform vendor, and the implementation case from a single agency without a control group. Both are directional. Import time windows quoted in industry write-ups conflict and need checking in your own account. This article reflects the position in July 2026 and requires quarterly review. It is not legal advice.