Skip to content

Conversion tracking · Google, Meta & LinkedIn

Conversion tracking for B2B lead generation: signals your ads can learn from, numbers your team can trust

Tracking has two jobs: give ad platforms a signal they can learn from in time, and give your team numbers they can trust. Zephra sends qualified-lead outcomes back to Google, Meta and LinkedIn where configured, checks they arrive, and uses them to decide what to change.

The job

What does conversion tracking need to do for B2B lead generation?

Short answer

It is recording what happens to a lead after the form, such as qualified, quoted or won, and matching that outcome back to the ad click. Google, Meta and LinkedIn can then report and, where eligible, optimise on those outcomes instead of treating every form submission as equally valuable.

Layer 1

Browser tag

Google tag, Meta Pixel or LinkedIn Insight Tag capture on-site actions and click identifiers. Necessary, but exposed to blockers, consent choices and browser limits.

Layer 2

Server-side events

The same events sent from your server or backend, with hashed contact details, so platforms still receive them when the browser doesn’t cooperate.

Layer 3

Sales outcomes

Qualified, opportunity, quote or won: stages the browser never sees, recorded by sales and sent back to the platforms against the original lead.

Capabilities

What Zephra tracks and sends

Zephra works from first-party and server-side conversion evidence and lead status from CRM data or a spreadsheet, and sends CRM lead stages to each platform through its official route, where configured.

PlatformHow Zephra sends lead stagesWhat the platform can do with them
Google AdsData Manager API, with enhanced conversions for leadsReport qualified and converted leads, and use them as bidding goals when volume and timing allow
Meta AdsConversions API with CRM lead stagesReport stages and, when eligible, optimise lead ads with Conversion Leads
LinkedIn AdsLinkedIn Conversions APIReport qualified leads against campaigns and use them for optimisation

Tracking on the product page →

Delivery

Did the events arrive? Delivery checks

Short answer

Check each platform’s diagnostics and track accepted, rejected and retried events over time. Google’s enhanced conversions for leads diagnostics rate data quality and flag missing identifiers; Meta Events Manager shows event activity and match quality. A silent failure makes bidding learn from incomplete data without any visible error.

Most tracking failures are not dramatic. A form is redesigned and the tag stops firing on one page. A CRM field changes name and uploads start failing validation. A thank-you page fires the Pixel and the server sends the same event without a shared ID, so Meta counts some leads twice. Reports still show conversions, just the wrong number of them.

That is why Zephra treats delivery as something to check continuously: events should be accepted, rejected events should be visible with a reason, and retries should not resend events that were already accepted.

Original framework

The B2B signal integrity scorecard

Eight checks. Score each green, amber or red, using the platforms’ own tools. Any red on checks 2, 3 or 8 is worth fixing before changing bids or budgets.

#CheckWhat good looks likeWhere to verify
1Identifier coverageMost leads carry a click ID (GCLID, fbclid, li_fat_id) or a hashed email or phoneYour form data or CRM export
2DeliveryEvents accepted; rejections rare and explained; no backlog of retriesGoogle enhanced conversions for leads diagnostics; Meta Events Manager; LinkedIn Campaign Manager conversions
3DeduplicationBrowser and server events share an event ID, so each conversion counts onceMeta Events Manager deduplication view; LinkedIn Insight Tag with Conversions API
4Match qualityMeta Event Match Quality rated Good or Great; no Google match warningsMeta Events Manager; Google Ads diagnostics
5Stage coverageSales records a status on nearly every leadCRM or spreadsheet
6Stage lag vs platform windowThe optimised stage lands inside the window: Google 63 days, Meta 28 days, LinkedIn’s recommended 90-day lead windowMedian days from lead to stage
7Consent and hashingConsent captured for measurement use; contact details normalised and SHA-256 hashed before sendingForm consent text; implementation review
8Goal hygieneOne primary optimisation action per platform; reporting-only actions marked secondaryGoogle Ads conversion goals; Meta and LinkedIn campaign settings

What changed

Offline conversions in 2026: Google, Meta and LinkedIn

Short answer

Google recommends enhanced conversions for leads: your site captures hashed details from the lead form, and qualified or won outcomes are uploaded later with the click ID where available. Since 15 June 2026 these uploads go through the Data Manager API, and outcomes must arrive within 63 days of the click.

PlatformRoute for offline outcomesTiming and volumeRecent change
Google AdsEnhanced conversions for leads through the Data Manager APIWithin 63 days of the click (90 for click-ID imports); a goal works best with 15+ conversions in 30 daysGoogle Ads API uploads blocked from 15 June 2026. Source
Meta AdsConversions API with CRM stages (Conversion Leads)200+ leads a month; stage within 28 days, reached by 1–40% of leads; uploads at least dailyThe Offline Conversions API is Meta’s legacy offline API. Source
LinkedIn AdsConversions API, deduplicated with the Insight Tag90-day click and view window recommended for lead conversionsQualified lead supported as a conversion type. Source

Platform rules change; check the linked documentation before relying on a figure. Last verified 15 September 2026.

Signals

Which stage should be the optimisation goal, and which is reporting only?

A deeper outcome is not automatically a better signal. A won deal that arrives four months after the click is outside every platform’s window, and four of them a month is too few to learn from. Send it for reporting, and steer on the earliest stage that sales agrees predicts value and that lands inside the window in enough volume, often a sales-accepted or qualified lead.

Mark everything else as secondary or reporting-only. Two primary goals on the same platform confuse bidding more than one imperfect goal does. The full argument, with worked numbers for industrial, SaaS and consulting funnels, is in the signal gap.

In practice

Recovered reporting is not new business

A UAE B2B software and IT services firm spends AED 80,000 a month on Google and Meta and gets 420 leads a month.

What the scorecard foundAct

Three measurement problems

Identifiers
65% of leads carry a click ID; 88% have an email.
Delivery
Google diagnostics show “needs attention”: the tag isn’t firing on one form.
Duplicates
9% of Meta website events are counted twice because the thank-you page sends no event ID.
After the fixesWait

Reported Google conversions rise 18%

What changed
Previously missed conversions are now counted. The number of qualified leads, from sales’ records, did not change.
The trap
Treating the jump as growth and raising budgets or targets on it.
Decision
Re-baseline targets, wait one conversion cycle, and judge campaigns on qualified leads before moving budget.

Getting started

No CRM? Start with a spreadsheet

You don’t need a CRM to start sending better signals. You need three things recorded consistently for every lead:

  1. An identifier captured with the lead: the click ID from the landing page URL and the email or phone number from the form.
  2. A stage that sales updates: accepted, qualified, opportunity or quote, won, lost.
  3. A date for each stage change, so you know whether it landed inside the platform’s window.

Zephra can work from CRM data or a spreadsheet. Phone and WhatsApp enquiries need the same treatment: record the outcome against an identifier you can match, or they stay invisible to the platforms.

Questions, answered

Conversion tracking FAQ

01Server-side tracking or pixel: which do I need?

Usually both. The browser tag captures on-site actions and identifiers; server-side events, such as the Meta Conversions API, add resilience and let you send outcomes the browser never sees, like a sales-qualified status. Use a shared event ID so the platform counts each conversion once rather than twice.

02Will better tracking increase my leads?

Not by itself. Fixing tracking usually increases reported conversions because previously missed events are now counted, but the number of real buyers hasn’t changed. Treat the jump as a measurement correction, re-baseline your targets, and judge campaigns on qualified leads before moving budget.

03Does Zephra send conversions to LinkedIn?

Yes. Where configured, Zephra sends CRM lead stages to LinkedIn through the LinkedIn Conversions API, as it does for Meta through the Conversions API and for Google through the Data Manager API.

04Does server-side tracking fix iOS and browser privacy losses?

Partly. Server-side events and hashed contact details improve matching when browsers block or limit tracking, but they don’t restore everything, and consent choices still apply. Be wary of anyone quoting a fixed recovery percentage.

05Is hashed data anonymous?

No. A hashed email can still be matched to a person, which is exactly why platforms accept it. Treat it as personal data and make sure consent and notices cover its use.

06What about phone and WhatsApp leads?

Record each call or chat as a lead with an identifier you can match, such as a phone number or a click ID from a tracked number, then update its stage like any form lead. Without that, those outcomes never reach the platforms.

07How are duplicate conversions prevented?

By giving browser and server events for the same action a shared event ID, which Meta and LinkedIn use to count it once, and by making sure retries never resend an event that was already accepted.

Start with your account

Find your measurement gaps before you change bids or budgets.

The free assessment checks what your conversions count, alongside wasted spend, structure and constraints, across Google, Meta and LinkedIn.

  • 10–15 minutes
  • No CRM required to start
  • No campaign changes during assessment