Measurement and advertising9 min read
How to measure the real return of your Google and Meta advertising
Platform conversions, CRM enquiries and booked revenue are different numbers. Here is how to define them and reconcile the three views.
- Author
- Félix Tarnóy
- Reviewed by
- Xavier Barlow
- Published
- Last reviewed
- Updated

Picture a Monday morning and a quarterly review. Google Ads reports 84 conversions for the month. Meta reports 61. The CRM holds 96 new enquiries. Finance can evidence six signed contracts. The figures are invented; the situation is not. No system is broken — they simply measure different things, on different calendars.
Advertising budget is rarely lost to weak creative. Far more often it is lost because the person approving the spend is relying on a number without knowing what is inside it. This article sets out how measurement is built from the click through to a qualified enquiry and on to booked revenue, and where the documented limits sit. You will not find step-by-step interface instructions here; the menus are rearranged every few months, the concepts are not.
The three numbers that will never agree
The sums are not small: Hungarian digital advertising spend reached HUF 231.8 billion in 2025, close to 56% of total domestic media spending. Real money, with real invoices behind it. The platform report, however, is not a ledger and was never built to be one.
Hungarian digital advertising spend reached HUF 231.8 billion in 2025, 11.2% more than in 2024, and close to 56% of total domestic media spending.
Hungary, calendar year 2025, digital advertising spend. Search advertising accounted for HUF 76.3 billion of the total.
IAB Hungary · 2026 · Source
The most consequential difference is the calendar. Google Ads books a conversion to the date the ad was shown, while most other tools attribute it to the date the conversion happened. The same documentation lists ten causes of discrepancy, notes that a conversion may still arrive up to 90 days after the click, and asks for 24 to 48 hours of synchronisation before figures are compared. Setting March spend against March revenue therefore compares two calendars.
On Meta’s side the attribution setting attached to each ad set decides what counts, and the reporting API returns 1-day, 7-day and 28-day click-based and view-based values in separate fields. A window existing as an API field does not mean it can be selected in the ad account: Meta’s attribution settings are mid-rollout. The same order can therefore appear in more than one window, depending on which one the person building the report pulled.
What the platform is actually counting
A conversion is not a fact of nature. It is a definition somebody typed into an account once. Four settings determine most of the reported number, and each of them can push a budget in the wrong direction within weeks.
The counting method
Google documents “every conversion” as the setting suited to tracking and improving sales, and “one conversion” as the choice where what matters is not the number of sales but whether a certain kind of lead was generated. In B2B this is the most common quiet error: an enquiry form left on “every conversion” counts again on every repeat submission, and the bidding then optimises towards that inflated figure. A change to the setting applies only to future conversions: you can stop the error, but you cannot correct it retrospectively.
The windows
The click-through conversion window defaults to 30 days and can be set from 1 to 30, 60 or 90 days depending on the source; the engaged-view window defaults to 3 days and the view-through window to 1 day. In a business with a three-month decision cycle, a 30-day window cannot see a large part of the result. That is not a fault. It is a setting somebody has to choose deliberately.
The attribution model
Google’s documentation currently lists two models: last click and data-driven, with data-driven the default for most conversion actions. First click, linear, time decay and position-based have gone. So if a report still works from those models, the channel splits inside it cannot be set against today’s.
| Setting | What it decides | Why it matters commercially |
|---|---|---|
| Counting method | Every conversion, or one per ad click | On lead generation, “every” inflates the count, and a correction only applies going forward. |
| Click-through window | How long after a click a conversion can still be booked to it | With a long decision cycle, a short window hides genuine results. |
| View-through window | Whether an impression without a click counts | It measures something other than search intent, so it belongs in its own column, not the headline total. |
| Attribution model | Last click or data-driven | The same order can be credited to a different channel, so a different campaign gets the budget. |
Consent is now a measurement input, not a legal formality
Consent mode works through several parameters, four of which carry the advertising and analytics signal: ad_storage, analytics_storage, ad_user_data and ad_personalization. They are not decoration on the cookie banner. They are inputs to the measurement. In basic mode tags are blocked until the visitor interacts with the banner and nothing reaches Google before that. In advanced mode they load with defaults set to denied, send cookieless measurement signals, and only that mode qualifies for the advertiser-specific model. The same website therefore reports different conversion counts depending on how its banner is configured.
Google’s position is unambiguous: advertisers must collect consent from end users based in the EEA and share those signals with Google in order to keep using the tags and SDKs behind measurement, personalisation and remarketing.
What turns this into a commercial risk is enforcement. Google states that reviewers visit a site as a user would and examine the information provided and the consents obtained. Google names eight audited issue types, including cookies set before consent, and warns that action on the account may follow, up to suspension of ad personalisation, remarketing and conversion measurement capabilities. Google allows a reasonable timeframe to put matters right, but publishes no fixed deadline.
Meta’s developer documentation points the same way: consent must be obtained before cookies or other trackers are set or read for purposes that are not strictly necessary, and the page names the GDPR and the ePrivacy Directive explicitly.
Free assessment
Does this sound familiar? Let us discuss the best place to begin.
Your assessment is carried out personally by one of our experts.
Modelled conversions: part of the number is an estimate
Both platforms say openly that what they cannot see, they estimate. Google states that modelled conversions use data that does not identify individual users to estimate the conversions it cannot observe directly. The documented gaps come from unconsented users, conversions that cross devices, browsers that restrict how long cookies survive, and ATT-impacted traffic. Processing can take up to five days to stabilise, which is why Monday’s figure can still move on Tuesday.
Modelling is not available to everyone. Google publishes a threshold of 700 ad clicks over a seven-day period, per country and domain grouping. The same page states that consented users are typically two to five times more likely to convert, and acknowledges that some conversions that genuinely occurred cannot be attributed to an ad click without cookies. Modelled conversions also appear inside the Conversions column, not broken out separately.
700 clicks / 7 days
In Analytics the bar is higher still: at least 1,000 events a day with analytics_storage set to denied for at least seven days, and at least 1,000 daily users with analytics_storage granted on at least seven of the previous 28 days — and Google adds that meeting those prerequisites does not by itself confer eligibility. It is worth stating plainly: the traffic on a typical Hungarian B2B site will not reach them. There, missing data does not reappear as an estimate. It is simply missing.
Server-side measurement: what it fixes, and what it does not
The tag running in the browser sees less every year, so both platforms push advertisers towards server-side measurement. That improves data quality. It does not remove the consent requirement, and it does not widen the windows.
Google: enhanced conversions
Enhanced conversions supplement the existing tags by sending hashed first-party customer data — email address, name, home address, phone number — hashed with SHA256 after normalisation and sent in hexadecimal form, and matched by Google to accounts the person was signed in to when they engaged with an ad. Translated into commercial terms: following the normalisation rules is not a technical footnote. The match rate, and therefore the quality of the measurement, depends on it.
Meta: the Conversions API
Meta describes the Conversions API as a connection between marketing data held on the advertiser’s server, website platform, app or CRM and Meta’s systems, with server events used for measurement and optimisation much as pixel events are. The pixel and the server feed run in parallel, which makes deduplication critical. Meta treats the two as a single event when the pixel’s eventID matches the server event’s event_id and the event names match — and only when the second event arrives within 48 hours of the first. Beyond that, the duplicate stays in the report.
The technical limits are strict. An event_time may be at most seven days in the past, and if a single event in a request breaches that, Meta returns an error for the whole request and processes none of it; for offline and physical store events the limit is 62 days. How well the customer information sent with an event can be matched to an account is reported through an Event Match Quality score out of ten, and that information must be hashed with SHA256 after normalisation — Meta states its systems are designed not to accept unhashed contact information at all.
The B2B ceiling is the import window, not the technology
This is where the platform reaches the end of what it can do alone. For importing offline conversions, Google Ads Data Manager reads back 90 days for file-based sources and 14 days for CRM and data-warehouse connections — the position in its documentation as at September 2026 — and names a click-to-conversion cycle shorter than 14 or 90 days as a prerequisite. On Meta’s side, offline events must be uploaded within 62 days.
If your sales cycle runs to five months — not unusual in manufacturing or any high-value purchase — then tracing the signed contract back to the ad inside the platform is structurally impossible. Not because it was configured badly, but because the import window is shorter than the decision. The loop then closes in the CRM rather than the ad account: send the platform an earlier, tightly defined signal, usually the qualified enquiry, and keep the truth about revenue in your own system.
Define the events in commercial language
Enquiry, qualified enquiry, proposal, order — each with a one-sentence written test. Until that exists, every other setting is guesswork.
Let the click identifier travel with the enquiry
The identifier arriving from the ad should pass through a hidden form field into the CRM and stay on the record. Without it there is nothing to join later.
Name one system as the source of revenue
The CRM or the invoicing system, not both. When two systems both claim to report revenue, the monthly review becomes an argument.
Send the platform an early signal, inside the import window
A qualified enquiry is settled within days; an order takes months. The algorithm can use the first. The second is what the board needs.
Reconcile three columns every month
Platform conversions, CRM enquiries, booked revenue — on one date logic. The goal is not that they match, but that the difference is explainable.
A worked scenario: the platform view against the ledger
| Line | Assumed value |
|---|---|
| Monthly media spend | HUF 1,200,000 |
| Conversions reported by Google Ads | 84 |
| Conversions reported by Meta | 61 |
| Unique enquiries recorded in the CRM | 96 |
| Of those, qualified enquiries | 24 |
| Of those, orders | 6 |
| Average contract value | HUF 900,000 |
| Contribution margin rate | 35% |
The two platforms together report 145 conversions, while the CRM holds 96 unique enquiries. The difference is not fraud: both can count the same person, on different windows, with part of the figure modelled. The usable metrics come from the CRM side. Cost per qualified enquiry is 1,200,000 / 24 = HUF 50,000, and cost per order is 1,200,000 / 6 = HUF 200,000. Contribution margin per order is 900,000 × 0.35 = HUF 315,000, so six orders contribute HUF 1,890,000 — more, in this hypothetical month, than the media spend.
It still does not follow that the budget should go up. Two questions are missing. The first is timing: if some of those six orders came from clicks in earlier months, the ratio mixes two periods. The second is incremental effect — how many of those customers would have found you anyway. No report answers that alone; it takes a controlled test, such as a deliberate pause or a geographic holdout. Reporting measures what happened. What would have happened otherwise can only be approached by experiment.
What to do next
None of the following needs a new platform or a new contract. It needs a few hours inside your own account, and one decision about which number the business will treat as authoritative.
- Ask which counting method your enquiry form conversion runs on. If this is lead generation set to “every conversion”, the bidding is optimising towards an inflated number.
- Check the click-through window against your genuine decision cycle. If the window is shorter, the report systematically understates campaigns that work.
- Pull one month of platform conversions alongside the enquiries in the CRM and count the unique records. The difference alone is decision-grade information.
- Have the cookie banner and consent signals tested as they actually behave, not as the settings screen describes them. Google’s audit looks through a visitor’s eyes.
- Confirm whether the click identifier reaches the CRM and stays on the record until the deal closes. This is the most common missing link.
- Write down, on one page, what a qualified enquiry and a customer are worth. That number decides what you may spend — not the platform’s recommendation.
If those answers do not come together in-house, the fix usually starts with measurement and landing pages rather than with the campaigns themselves. That work is described under Google Ads and Meta ads management, and on the revenue side it is often custom software and CRM that closes the loop. To begin with an outside pair of eyes, the free digital assessment covers the advertising account and its measurement as a separate module.
Sources
IAB Hungary — AdEx 2025 and AdExpect 2026
Hungarian digital advertising spend reached HUF 231.8 billion in 2025, 11.2% more than in 2024, and close to 56% of total domestic media spending.
Hungary, calendar year 2025, digital advertising spend. Search advertising accounted for HUF 76.3 billion of the total.
https://iab.hu/2026/03/27/232-milliard-forintos-digitalis-reklampiac-szerenyebb-idei-varakozasok-megjelent-az-adex-2025-es-az-adexpect-2026/Last reviewed: 1 September 2026
Google Ads Help — Why Google Ads conversion numbers differ from other reporting tools
Google documents that Google Ads reports conversions on the ad impression date while other reporting tools attribute them to the conversion date, lists ten factors behind the resulting discrepancies, states that conversions may still arrive up to 90 days after the click, that conversions detected as spam or invalid traffic are removed after the fact, and asks for 24–48 hours for data synchronisation.
https://support.google.com/google-ads/answer/7457111Last reviewed: 1 September 2026
Meta for Developers — ads-action-stats — attribution windows in reporting
Meta’s reporting API returns 1-day, 7-day and 28-day click-based and view-based values in separate fields (1d_click, 1d_view, 1d_ev, 7d_click, 7d_view, 28d_click, 28d_view), alongside values for the data-driven model (dda), incrementality and the inline attribution window.
https://developers.facebook.com/docs/marketing-api/reference/ads-action-stats/Last reviewed: 1 September 2026
Google Ads Help — Conversion counting options
Google documents “Every conversion” as suited to tracking and improving sales, and “One conversion” as the choice when what matters is whether a certain kind of lead was generated rather than the number of sales; the setting is reflected in the Conversions, All conversions and view-through conversion columns, and any change applies only to future conversion reporting.
https://support.google.com/google-ads/answer/3438531Last reviewed: 1 September 2026
Google Ads Help — Conversion windows in Google Ads
The click-through conversion window defaults to 30 days and can be set from 1 to 30, 60 or 90 days for Search and Display campaigns depending on the conversion source; the engaged-view window defaults to 3 days, the view-through window to 1 day, and overall the window can be set anywhere from 1 to 90 days depending on the source.
https://support.google.com/google-ads/answer/3123169Last reviewed: 1 September 2026
Google Ads Help — Attribution models in Google Ads
Google Ads currently offers two attribution models, last click and data-driven; data-driven is the default for most conversion actions, and the first click, linear, time decay and position-based models are no longer supported.
https://support.google.com/google-ads/answer/6259715Last reviewed: 1 September 2026
Google for Developers — Consent mode: consent types and modes
Google consent mode uses, among others, the ad_storage, analytics_storage, ad_user_data and ad_personalization parameters; in basic mode tags are blocked until the user interacts with a consent banner and no data is sent before consent, in advanced mode tags load with defaults set to denied and send cookieless measurement signals, and only advanced mode qualifies for the advertiser-specific model.
https://developers.google.com/tag-platform/security/concepts/consent-modeLast reviewed: 1 September 2026
Google Ads Help — EEA end-user consent requirement
Google states that advertisers must collect consent for the use of personal data from end users based in the EEA and share consent signals with Google in order to keep using the applicable tags and SDKs for measurement, ad personalisation and remarketing; where the two parameters are set with conflicting values, personalisation is disabled.
https://support.google.com/google-ads/answer/13695607Last reviewed: 1 September 2026
Google Ads Help — EU user consent policy audits
Google states that a team of reviewers visits a website or app as a user would and examines the information provided and the consents obtained; it names eight audited issue types, including cookies set before consent and an incorrectly configured consent signal, and warns that action on the account may follow, including suspension of audience functionalities such as ad personalisation and remarketing and of conversion measurement capabilities.
https://support.google.com/google-ads/answer/16724512Last reviewed: 1 September 2026
Meta for Developers — Cookie consent resource
Meta’s developer documentation states that user consent must be obtained before setting or reading cookies or other trackers for purposes that are not strictly necessary or otherwise exempt, naming European laws such as the GDPR and the ePrivacy Directive, and describes valid consent as freely given, specific and informed, and unambiguous and affirmative.
https://developers.facebook.com/docs/privacy/Last reviewed: 1 September 2026
Meta for Developers — Data processing options — Limited Data Use
Meta describes Limited Data Use as a data processing option supporting compliance with various US state privacy regulations, lists the states covered, and warns that enabling it may affect campaign performance and effectiveness, limit retargeting and measurement capabilities, and affect audience size.
https://developers.facebook.com/docs/marketing-apis/data-processing-optionsLast reviewed: 1 September 2026
European Commission — Legal grounds for processing personal data
A controller must have — and document — a valid legal basis for processing personal data.
https://commission.europa.eu/law/law-topic/data-protection/information-business-and-organisations/legal-grounds-processing-data_enLast reviewed: 1 September 2026
Google Ads Help — Modelled online conversions
Google states that modelled conversions use data that does not identify individual users to estimate conversions Google is unable to observe directly; documented gaps include unconsented users, cross-device conversions, browsers that restrict cookie measurement and ATT-impacted traffic, and modelled conversions can take up to 5 days to fully process and stabilise in reporting.
https://support.google.com/google-ads/answer/10081327Last reviewed: 1 September 2026
Google Ads Help — Conversion modelling with consent mode
Google documents a threshold of 700 ad clicks over a 7-day period, per country and domain grouping, for consent mode conversion modelling; states that consented users are typically 2–5 times more likely to convert than unconsented users; places modelled conversions inside the Conversions column; and acknowledges that some conversions that in reality occurred may not be accounted for, being unattributable to ad clicks without cookies.
https://support.google.com/google-ads/answer/10548233Last reviewed: 1 September 2026
Google Analytics Help — Behavioural modelling for consent mode (GA4)
Google Analytics behavioural modelling requires a property to collect at least 1,000 events per day with analytics_storage set to denied for at least 7 days, and to have at least 1,000 daily users sending events with analytics_storage granted on at least 7 of the previous 28 days; Google states that meeting these prerequisites does not guarantee eligibility.
https://support.google.com/analytics/answer/11161109Last reviewed: 1 September 2026
Google Ads Help — Enhanced conversions
Enhanced conversions supplement existing tags by sending hashed first-party customer data — email address, name, home address and/or phone number — hashed with hex SHA256 after normalisation, and matched by Google to accounts users were signed in to when they engaged with an ad; Google states that it reports only privacy-safe conversions.
https://support.google.com/google-ads/answer/9888656Last reviewed: 1 September 2026
Meta for Developers — Conversions API
Meta describes the Conversions API as a connection between an advertiser’s marketing data — from a server, website platform, app or CRM — and Meta’s systems, covering website, app, business messaging and offline conversion events, with server events used in measurement, reporting and optimisation in a similar way to other connection channels such as the pixel.
https://developers.facebook.com/docs/marketing-api/conversions-apiLast reviewed: 1 September 2026
Meta for Developers — Deduplicate pixel and server events
Meta states that deduplication requires the pixel’s eventID to match the Conversions API event_id and the event names to match; Meta generally prefers the event received first, and events are only deduplicated if they are received within 48 hours of the first event carrying that event_id.
https://developers.facebook.com/docs/marketing-api/conversions-api/deduplicate-pixel-and-server-eventsLast reviewed: 1 September 2026
Meta for Developers — Using the Conversions API — limits
Meta documents that event_time may be at most 7 days before an event is sent, and that if any event_time in the request exceeds this Meta returns an error for the entire request and processes no events; for offline and physical store events transactions should be uploaded within 62 days of the conversion, and up to 1,000 events may be sent in a single request.
https://developers.facebook.com/docs/marketing-api/conversions-api/using-the-apiLast reviewed: 1 September 2026
Meta for Developers — Conversions API best practices and Event Match Quality
Meta states that an event’s Event Match Quality score, out of 10, indicates how effective the customer information sent with a server event is at matching it to a Facebook or Meta account; action_source is required for all events, event_source_url and client_user_agent for all website events, external_id and event_id are recommended, and events carrying only sparse parameter combinations are rejected.
https://developers.facebook.com/docs/marketing-api/conversions-api/best-practicesLast reviewed: 1 September 2026
Meta for Developers — Customer information parameters and hashing rules
Meta requires customer information parameters — email, phone, first and last name, date of birth, gender, city, state, postcode and country — to be hashed with SHA256 after normalisation, and states that its systems are designed not to accept unhashed contact information, which is used for matching purposes only.
https://developers.facebook.com/docs/marketing-api/conversions-api/parameters/customer-information-parametersLast reviewed: 1 September 2026
Google Ads Help — Importing offline conversions — import windows
Google Ads Data Manager imports conversions from 90 days ago on every run for Google Cloud Storage, Amazon S3, HTTP, SFTP and Google Sheets, imports the last 14 days for Salesforce and HubSpot and for BigQuery, Redshift, Snowflake, MySQL and PostgreSQL, and names a click-to-conversion cycle shorter than 14 or 90 days depending on the source as a prerequisite.
https://support.google.com/google-ads/answer/7012522Last reviewed: 1 September 2026