How to send qualified leads back to Google Ads
To send qualified leads back to Google Ads, your CRM must record a confirmed business milestone as a separate event and send it to the appropriate conversion action. A submitted form alone does not establish that milestone. You need a documented qualification decision, the time it actually occurred and a link to the original enquiry. You can then check processing, attribution and how campaigns use the signal.
Reliable lead feedback starts with sales
Google distinguishes between the goals Qualified lead and Converted lead. Converted lead refers to a completed step in your sales process that you choose to measure; it does not necessarily mean a paid order. These labels do not replace your own business definition. An enquiry that meets a consultancy's qualification criteria might still require preliminary assessment at a technical services company. First define the verifiable event that your business will recognise as a qualified lead.
This guide turns that decision into a practical feedback process. Its central working document is the lead feedback plan: it records who qualifies a lead, which criteria they apply, what event the decision creates, where it is sent and how discrepancies are handled. A technical connection alone does not complete this process.
Updated: 12 September 2026. The aim is an auditable process connecting advertising and sales. Acceptance of an upload proves neither attribution to an ad nor additional revenue.
What counts as a qualified lead in your business?
Write criteria that two people will apply consistently
A useful definition describes observable characteristics. For example, the enquiry concerns a service you offer, comes from an area you serve, has a reachable business contact and expresses a specific need. “Interesting” or “a good impression” is not a sufficient criterion. The person reviewing the lead must be able to tell when a condition is met and what information is still missing.
Separate essential requirements from information that becomes available later. An unknown project budget may remain an open question without making the enquiry worthless. A personal enquiry about an exclusively B2B service, however, may clearly fall outside your remit. Record reasons for rejection as well, so that unsuitable enquiries and cases awaiting review do not share the same status.
The Google guidance on importing different sales stages recommends separate conversion actions for distinct events. Having several measurable stages does not mean they should all become bidding goals at the same time. Deliberately choose the stage that reflects your current quality objective reliably and early enough to be useful.
Write the definition as one sentence in your lead feedback plan and add three borderline cases: a suitable enquiry without an appointment, a reachable contact whose request falls outside your services, and a suitable prospect whose project starts later. If sales and marketing assess these cases differently, resolve the definition before approving the technical setup.
Separate CRM status, feedback events and advertising goals
A status shows the current position; an event preserves the history
A CRM field such as “qualified” often shows only the current state. Reliable feedback also needs the history: when qualification was first confirmed, who confirmed it and what evidence supported the decision. If a later status overwrites this information, the original feedback becomes difficult to audit. Store the stage event separately from the changing operational status.
One contact may submit several enquiries, and one enquiry may involve several contacts. Decide whether your unit of measurement is the contact, the specific enquiry or the sales opportunity. For many B2B processes, the individual opportunity is easier to audit. Record this decision in the plan before linking identifiers from different systems.
Assign responsibility within sales for business qualification, nominate an owner for the data flow and make someone in marketing responsible for conversion actions and campaign goals. In a small business, one person can hold several roles. The responsibilities should still remain distinct, so nobody mistakes a successful import for approval of the underlying qualification decision.
If you have not yet defined status transitions, cover arrangements or callback rules, first organise CRM statuses, responsibilities and response times. Otherwise, automation will simply transmit inconsistent sales decisions more quickly.
Complete the lead feedback plan
One row per event type and an auditable trail for each case
The plan consists of a rules overview and a related event log. The overview explains how a stage works. For each individual case, the log records the event identifier, event time, transmission attempt and result. Keep these two levels distinct: a clearly documented rule does not prove that every relevant record was actually sent.
| Plan field | Decision | Evidence | Owner | Check | Reason to stop |
|---|---|---|---|---|---|
| Qualification | Business need and service align; appropriate contact identified | Documented conversation note | Sales | Borderline cases assessed consistently | Subjective judgement only |
| Source event | First confirmed qualification per opportunity | Stage history with actual event time | CRM owner | Identifier and time remain stable | Only current status is available |
| Feedback | One approved transmission method | Attempt log with result | Data owner | Errors and retries are traceable | Uncontrolled parallel imports |
| Goal and report | Named action in the correct account | Action ID, category and goal usage | Google Ads owner | Processing distinguished from attribution | Unclear campaign goal |
| Correction | Distinguish an incorrect report from a later lost deal | Correction record with a reason | Sales and marketing | Supported correction method confirmed | Date or identifier is overwritten |
Also record the definition version, approval date, deputy and location of supporting evidence. Use internal references in the shared working document instead of unnecessary personal information in free-text fields. The responsible people must be able to review the original case without duplicating customer data across additional spreadsheets.
From a new enquiry to confirmed qualification
A fictional B2B service provider makes the distinctions clear
The fictional company Rheinblick IT serves small businesses in Hesse. On 2 September, it receives an enquiry about a workplace IT project. The form creates opportunity RB-1042. During a conversation on 4 September, sales confirms the business need, service fit and an appropriate contact person. Only this conversation satisfies the approved definition; the form submission remains a separate, earlier event.
The CRM therefore records a qualification event for RB-1042 at the actual time the conversation's outcome was confirmed. The event is sent in the next scheduled run. The run time does not replace the event time. If a contract is signed later, that would create another business event, rather than a second copy of the qualification.
The Google instructions for setting up offline conversions from clicks describe the connection between an ad interaction, lead capture and a later offline outcome. To operate the process, you also need your own evidence trail: the same case must remain traceable from the enquiry through the decision to the feedback sent.
An internal log entry could therefore contain opportunity RB-1042, the event “first qualified”, the event time “4 September 2026, 10:15, UTC+02:00”, the definition version and the responsible person. Add the reference for the original enquiry contact and the transmission status. These details describe your internal operating model; they are not a universal upload template. The transmission method determines which fields actually go to Google and in what format.
If sales reopens RB-1042 after a callback, that does not create a new first-qualification time. If a separate new project follows later, your business rule must determine whether it constitutes a new opportunity. That decision must not depend by chance on whether the CRM happened to create a new record.
Recognise the same event across retries
Use three layers of protection together
First, your system needs a stable internal event identity, such as a combination of opportunity, stage and an event version justified by the business process. Second, every transmission attempt must reference that event. Third, correctly populate the fields supported by your chosen import method for recognising previously submitted events. An internal identifier does not automatically deduplicate events in Google Ads.
The conversion counting setting does not solve this problem on its own either. “One” counts at most one conversion per ad interaction for the relevant action; it does not mean one lead per person across their entire contact history. It replaces neither your business rule nor the technical handling of repeated uploads. In particular, deduplication does not span two different conversion actions merely because they describe related activity.
At Rheinblick IT, a retry therefore still references the same first qualification of RB-1042. The system does not generate a new event time or a new qualification decision simply because the first attempt returned an error. It logs the attempt and its result while preserving the historical decision.
Withhold approval if the same stage is being sent through a CRM integration, file upload and another automated method without controls to prevent duplication. Check the event recognition features actually supported by your combination of import method and conversion action.
Send only documented events with usable attribution data
A contact detail proves neither an event nor permission to use it
Feedback requires the information that your import method specifies for attribution, the event and the destination. Check these requirements against the relevant data specification. A populated email field proves neither successful attribution nor the suitability of an enquiry. Only send data whose use for this particular purpose has been established within your business.
Technical collection and validation of attribution data are covered in the guide to setting up and checking Enhanced Conversions for web and lead data. The focus here is the responsibility that follows: sales must be able to substantiate the milestone reported, and the data process must not fill an evidence gap with a default value.
Observe the current import requirements and time limits: uploads are no longer accepted more than 90 days after the associated last click for traditional offline conversions, or more than 63 days after that click for enhanced conversions for leads. Also check the configured conversion window and the import history selected in the source system. These limits apply to different parts of the process. They are not a recommendation to delay feedback.
If an actual event can no longer be submitted within the applicable limits, preserve it accurately in the CRM and document why it is ineligible for feedback. Never artificially move it to a more recent date. The lead feedback plan should show which qualification decisions exist and which of them can potentially contribute to advertising measurement.
Organise conversion actions around business stages
Explicitly approve both measurement and bidding use
Each stage needs an unambiguous destination action. Record the account, action name, technical reference, category and intended use in campaigns. A clear name helps people, but does not replace checking the action actually connected. The following overview is one possible working arrangement, rather than a default configuration suitable for every account.
| Stage | Evidence | Separate action | Initial use | Approval question | Typical mistake |
|---|---|---|---|---|---|
| Enquiry received | Recorded case | Form enquiry | Review existing measurement | What is currently being optimised? | Treating a received enquiry as a qualified lead |
| Qualified | Confirmed criteria | Qualified lead | Initially monitor under controlled conditions | Are the definition and feedback stable? | Sending an unchecked status change |
| Completed step | Evidence of the chosen later outcome | Converted lead | Measure the distinct later outcome | Is there sufficient evidence for the chosen step? | Renaming qualification as another outcome |
| Lost | Outcome with a recorded reason | No positive qualification event | Analyse in the CRM | Was the earlier qualification correct? | Treating every loss as a measurement error |
Check the rules for primary and secondary conversion actions. With standard goals, bidding requires a primary action and a goal that the campaign uses. Secondary actions normally serve an observation role, but they can still be used for bidding when included in a custom goal. “Secondary” alone is therefore not sufficient protection for an observation-only test.
Confirm the actual campaign configuration in the plan. The decision to make a quality stage a bidding goal comes only after business and technical approval. Successfully creating the action does not make that decision automatically.
Set up lead feedback step by step
From a confirmed status transition to an approved goal
- Have sales confirm qualification and attach the specific supporting evidence to the case.
- Create exactly the approved stage event, with a stable reference and its actual occurrence time.
- Check required information, attribution data and feedback eligibility before adding the case to the transmission queue.
- Map the event to the approved conversion action in the correct account.
- Send it through the agreed method and save the result for each event wherever that method provides one.
- Assign an owner and a next review date to errors and unresolved cases.
The companion guide explains how to assign primary, secondary and account-default conversion goals correctly. Your lead feedback plan also defines when a record meets the business requirements for sending. A scheduled import must not select an unreviewed case simply because a field has been populated.
Next, check the use of account-default and campaign-specific conversion goals. Document which stage currently counts for each affected campaign. Any change to this mapping needs its own entry in the plan's change log. This keeps a technical extension of the data flow distinguishable from a change in what the campaign is intended to achieve.
Manage retries, outages and integration changes
Give the transmission process a clear operational owner
Define what happens after a failed run: which events are confirmed as processed, which were rejected and which remain unresolved? Rerunning every record indiscriminately can introduce further uncertainty. Retry using the original event data and the retry rules of the system involved. Keep an unresolved attempt explicitly marked as unresolved.
For custom development, the current Google documentation for uploading offline conversions is the authoritative reference. Before implementing a new integration, check which API route is available for your access. The continued availability of an older method, or an example from an existing project, does not establish that a new integration can use the same route.
When changing transmission methods, define a traceable cutover: which events remain the old method's responsibility and where the new method takes over. Any trial period in which the methods overlap requires demonstrably controlled handling of the same events. Simply hoping that automatic deduplication will deal with them is not enough.
Rheinblick IT therefore keeps an open error log containing the event reference, latest response, responsible person and next action. An interrupted import triggers technical investigation. It does not require sales to mark the affected leads as qualified again.
Validate the process with traceable real cases
Include difficult cases in the sample
Start with existing cases that you are entitled to use and check whether the rules work. Include an enquiry that genuinely qualifies, one still awaiting review, a rejected case and one that has been reopened. Do not send invented successful conversions to a live advertising account as supposed test data. Technical prechecks also cannot replace reviewing the underlying business events.
For RB-1042, the sample must trace the CRM evidence through to the correct action. The unresolved comparison case must not yet produce a qualification event. The reopened case must not create an unjustified second qualification. These checks reveal different problems from simply checking whether all required fields are populated.
The offline data diagnostics help investigate data quality and processing issues. Read the specific status and identify the layer it concerns. A diagnostics screen with no apparent issues does not automatically confirm that your business definition is meaningful or that every CRM case could be attributed to an ad interaction.
For each sampled case, explicitly document the expected result before looking at the technical status. If you decide what counts as correct only after the import, selection errors can easily go unnoticed. Keep the expectation, actual observation and resulting correction together. The next reviewer can then see whether the same problem has returned.
Approval requires two signatures or recorded confirmations: sales confirms the selected events, and the data owner confirms their technical handling. Marketing additionally checks goal usage. In a small team, one person may perform several checks; the log must still show them as separate decisions.
Reconcile CRM events, transmission and Google Ads reports separately
Different totals may count different things
Within the CRM, use a stable business cohort and event type. For Ads, align the event type, reporting period and report settings; an aggregate Ads report does not automatically reproduce your CRM cohort of newly created enquiries. The figures pass through several layers, and each needs its own denominator. Count unique events for reconciliation and report the number of technical transmission attempts separately.
| Layer | What is counted? | Reference | Possible gap | Check | Owner |
|---|---|---|---|---|---|
| Confirmed in CRM | Actual qualification events | Stage history | Incomplete review | Definition and event time | Sales |
| Eligible for feedback | Events meeting usable input requirements | Approved selection | Missing attribution data | Exclusion reasons documented | Data owner |
| Sent | Unique events with a transmission attempt | Import log | Outage or selection error | Case linked to its attempt | Technical operations |
| Processed | Events confirmed as processed | Processing result | Errors or processing still pending | Diagnostics and required waiting time | Technical operations |
| Attributed in Ads | Reported, attributed conversions | Appropriate action and report column | Attribution, time basis or report scope | Comparable settings | Marketing |
Google explains possible conversion measurement discrepancies. For your reconciliation, record the account, action, date range, time zone, counting method and column used. Standard reports generally assign conversions to the time of the ad interaction. To reconcile events, you could use “All conv. (by conv. time)”, segment by conversion action and align the time zone. Compare processed results using their actual conversion time; the upload date is not a substitute.
In the teaching example, eight of nine confirmed qualifications might be eligible for feedback. If eight are sent, seven are confirmed as processed and six are attributed in the appropriate Ads report, there are initially three separate questions: why was one case excluded before transmission, why is one processing result unresolved, and what explains the reporting difference? Keep these investigations separate. Do not infer individually proven attribution from aggregate figures when the report itself does not provide it.
Do not demand artificial agreement between CRM and Ads totals. Require a reasoned explanation of the differences. Record an unexplained gap in the plan with an owner and a review date; renaming a metric does not resolve it.
Assess quality using enquiry cohorts with enough follow-up time
Leads qualified later change the picture of an earlier period
For business analysis, group enquiries into cohorts by when they arrived. Then observe the proportions reviewed, qualified, still open or completed. The qualification rate needs an explicit denominator: all incoming enquiries tell a different story from only those that have already been reviewed.
Consider a deliberately simplified calculation. Of 20 enquiries in a fictional intake cohort, 16 have been reviewed. Of those, nine qualified and seven were rejected; four remain unreviewed. Nine out of 20 is 45 per cent of all enquiries, while nine out of 16 is 56.25 per cent of reviewed enquiries. Both calculations are correct, but they answer different questions. These figures are a teaching example, not a campaign performance benchmark.
At Rheinblick IT, some enquiries remain open for several days because the relevant contacts cannot be reached immediately. Comparing this recent week with an older week in which every case has been reviewed would be premature. Determine the necessary follow-up period from your actual review and decision times rather than adopting an unsupported universal timeframe.
Advertising reports can also be affected by conversion delays as results are added later. Distinguish three waiting periods: progress through the sales process, transmission and processing, and the timing of attributed results in reports. A technical backlog requires a different remedy from a slow sales process.
Then examine the connection to the business: are qualified enquiries actually followed up, do suitable projects result, and do rejection reasons remain traceable? A rise in reported qualified outcomes may reflect more complete measurement. On its own, it does not prove that advertising generated additional suitable prospects.
Correct inaccurate feedback and interpret later losses properly
Losing a deal does not automatically invalidate its earlier qualification
RB-1042 may have been correctly qualified and later choose not to work with Rheinblick IT. The later outcome does not necessarily change the truth of the earlier event. Record the loss and its reason in the CRM, and assess the later sales stage. Do not delete the historical qualification solely because it did not lead to an order.
The situation is different if the enquiry was mistakenly marked as suitable or the same event was improperly reported more than once. The plan then needs a correction case containing the original event reference, affected action, actual error, owner and available correction method. Distinguish an incorrect original report from a subsequent change in the opportunity.
The Google rules for conversion adjustments distinguish between retractions and restatements, among other adjustments. A restatement to a non-zero value changes the value; a retraction removes the counted conversion. A restatement to zero can also retract it. Check which adjustment is possible, and which identifiers and time limits apply, for the specific conversion and transmission method. Changing a CRM status does not automatically correct an import that has already been processed.
A retraction of the original record cannot simply be reversed through another adjustment. Google documents a separate recovery procedure involving a new import for certain offline cases. Treat this only as a separately reviewed repair, not as a routine retry or a new business qualification. Avoid improvised offsetting events, negative placeholders or new timestamps. If your method cannot correct a case, document that limitation, its effect on reporting and the intended handling. Then fix the cause in the event-generation process so that the same type of incorrect feedback does not recur.
When to pause feedback or its use in bidding
Match the response to the affected layer
Unclear definition
If different employees assess the same case differently, correct the business rule. Technical optimisation will not make those events more reliable.
Damaged history
If genuine event times are missing or references are recreated on retries, stop the affected submissions until the historical records can be reconciled.
Interrupted data flow
When a backlog develops, preserve unresolved events and address the import error. Do not change the qualification definition at the same time simply to make reports agree.
Unchecked goal usage
If it is unclear which campaigns use the new action, do not approve a planned observation phase as risk-free. Check custom goals as well.
“Stop” does not always mean turning off all advertising. Specify whether to hold individual events, interrupt a transmission method or postpone a planned change to bidding goals. The treatment of existing events with reliable measurement remains a separate decision.
Every pause needs an owner, a specific repair and a condition for approval to resume. Without these, a prudent pause can turn into a permanently neglected backlog. The lead feedback plan keeps unresolved cases visible and prevents a later successful run from concealing earlier errors.
Organise approval into verifiable steps
Evidence determines when to move to the next stage
Use a plan with verifiable conditions rather than a blanket promise to finish after a fixed number of days. Small datasets, long sales cycles or incomplete histories may extend the assessment. Begin the next step when the evidence required for it is available.
| Step | Result | Evidence for approval | Owner | Next decision |
|---|---|---|---|---|
| Definition | Shared qualification rule | Borderline cases resolved consistently | Sales | Set up event generation |
| Technical review | Traceable events and attempts | Sample checked, including a retry | Data owner | Allow controlled observation |
| Reconciliation | Explained differences between layers | Errors addressed and cohort maturity considered | Marketing and sales | Assess goal usage separately |
| Ongoing operation | Accountable feedback process | Cover arrangements, correction method and change log in place | Named process owner | Review again when changes occur |
During routine operation, monitor more than error messages. Check whether the proportion of unreviewed enquiries is growing, whether rejection reasons are shifting and whether new employees apply the same definition. An import can run without errors while sending poorer business data if the sales process changes unnoticed.
Agree a regular review schedule that reflects actual enquiry volume and the importance of the signal. A deputy must know where unresolved cases are recorded and who can grant approval. For the next review, have at least the oldest unresolved events, unexplained error messages and rules changed since the previous review ready. This keeps the feedback process operational even when the person who originally set it up is unavailable.
Version changes to the definition, event generation and goal usage separately. This makes it possible to investigate later whether a change in the report could stem from new demand, different qualification criteria or a technical change.
FAQ: Google Ads and qualified leads
Can I simply import every form enquiry as a qualified lead?
Only if submitting the form fully and demonstrably satisfies your business qualification criteria. Many B2B processes still lack information at that point. A new label does not improve enquiry quality. Measure receipt as its own stage and report the later confirmed qualification separately.
Do I need a large CRM system to send lead feedback?
The essential requirements are reliable event history, clear responsibilities and a supported transmission method. A simple process can meet these requirements too. A basic spreadsheet becomes problematic, however, once timestamps are overwritten, retries go unrecognised or unresolved errors stop receiving attention.
Should I use Qualified lead or Converted lead?
Choose the category that matches the actual event. Confirmed suitability is different from a completed step that you choose to measure in the sales process. That later step does not necessarily mean a paid order. Both stages may be worth measuring. Deciding which should become a campaign goal is a separate question, influenced in part by reliability and how promptly the signal becomes available.
Does the “One” counting option prevent all duplicate leads?
No. It governs counting within the relevant conversion action under the platform's rules. It replaces neither a stable internal event identity nor checks on repeated transmissions. It also does not automatically combine separate actions into a single business-level lead count.
Why does my CRM show more qualified leads than Google Ads?
The CRM can include cases without usable attribution to an ad. Other differences arise from feedback requirements, processing, the time basis, counting settings and report scope. Review the CRM cohort internally and align event type, time basis and scope for the Ads comparison. The difference alone does not reveal where an error occurred.
Must I retract qualification if the lead is later lost?
Not automatically. If qualification was correct at the time, it remains a valid historical event. The later loss is a separate outcome. An incorrect original submission, however, needs a justified correction using an adjustment method supported for the affected conversion.
How soon can I use the new quality goal for bidding?
This guide cannot provide a fixed date for approval. Check the definition, retries, processing, goal configuration and data with enough time to mature. A small successful import provides technical evidence; it is not a complete basis for changing how campaigns are managed.
Does a higher number of qualified conversions prove advertising is working better?
No. More reported conversions may also result from more complete measurement or changed criteria. Review comparable cohorts, sales follow-up and subsequent business outcomes. Establishing whether advertising caused additional demand requires more than comparing two reported conversion totals.
The value of feedback depends on its quality
Start with a verifiable event and a named owner
A useful feedback process connects the enquiry to confirmed qualification, technical transmission and a checked report. Each transition needs its own evidence. That helps you identify whether an unusual figure originates in sales, data processing or report interpretation.
Begin with one clearly defined stage and a limited, traceable set of cases. Complete the lead feedback plan, check borderline cases and retries, and explain the differences between CRM and Ads. Then assess whether to change campaign goals. The value comes from reliable quality, rather than from reporting as many different statuses as possible.
Would you like to examine the connection between sales quality and advertising measurement? Arrange a review of Google Ads and CRM feedback with Salestudia. The focus is a clear qualification definition, traceable events and controlled use of the results.