For businesses spending $100K+ per month in ad spend
LGG Media, an Ottawa-based agency, manages Meta campaigns for businesses spending $100,000 or more per month. Every paid social account runs on the same offline conversion infrastructure as our paid search accounts, real revenue events passed via the Conversions API, not platform-estimated conversions. Wondering how Meta stacks up against Search for lead generation? Google Ads vs Facebook Ads, answered with portfolio data.
Meta's default attribution relies on pixel events. But many conversion journeys continue offline, a phone call, a consultation, a signed contract, a collected payment. The pixel gives Meta's algorithm an incomplete picture of who actually became a customer, so budget spreads across revenue and non-revenue traffic at equal weight.
When the pixel is the only signal, Meta's audience models form around the top of the funnel, where the visible activity is. Segments that produce paying customers and segments that produce only inquiries get equal budget weight, so cost per acquired customer rises while reported conversion volume holds steady.
Meta reports conversions on its attribution windows; the CRM records what actually closed. On offline-funnel accounts the gap regularly exceeds 40 percent. Bidding from platform figures understates the true cost of a customer and scales toward audiences that generate activity, not revenue.
40%+, typical discrepancy between Meta-reported conversions and CRM-recorded revenue on offline-funnel accounts.
Every Meta account we manage runs on the same offline conversion infrastructure as our paid search accounts. Revenue flows from the client's CRM and payment processor, through the tracking layer, to Meta via the Conversions API with real transaction values attached. The bidding target is set on actual revenue.
Advantage+ and manual structures. We use Advantage+ where verified data volume supports broad optimization, and manual structures where the business type or industry restriction requires tighter control. The choice comes from the attribution data gathered during the engagement.
Pixel and CAPI deduplication. When both the pixel and the Conversions API are active, the same conversion can be sent twice. Our data layer deduplicates before events are pushed through CAPI, using the event ID to stop Meta from counting the same revenue event twice.
Lookalike audiences are built from Hyros lead records, filtered to verified high-value customers. Source-audience quality depends on the accuracy of the revenue data behind it, and each new verified event improves the model over the engagement.
Seeded from your highest-paying customers, giving Meta a signal weighted toward revenue, not volume. The algorithm finds users who resemble your highest-value customers.
Seeded from your highest-LTV customers, pointing spend toward long-term revenue rather than one-time transactions. The model compounds as LTV data accumulates.
Customers who converted from specific campaign types are isolated as separate seeds, so Meta builds distinct models for search-initiated and social-initiated customers.
Each account ran Meta on the same offline conversion infrastructure as its Google campaigns, revenue-based bidding on verified CRM data, not platform-reported estimates.
Monthly premium value per enrolled policy passed to Google and Meta at once. Attribution held at 92% against CRM revenue as spend scaled from $150,000 to $3.9 million per month. Google published a 2025 case study documenting a 23% revenue increase in 45 days.
Confirmed booking value passed from the CRM to both platforms gave Meta the same signal as Google. The unified signal scaled direct-booking segments across both channels and ended dependence on lead aggregators. Attribution accuracy: 93%.
Signed retainer value by case type pushed from Salesforce to Google and Meta as the offline conversion event. A consistent signal across both platforms kept spend on the highest-retainer segments for four consecutive years.
Meta performs best on the accounts where both Google and Meta receive the same offline conversion signal. When both platforms work from verified CRM revenue, their audience models form around the same definition of a high-value customer.
When both platforms get accurate offline data, each builds an accurate model of a high-value customer from its own data set. Traffic quality improves on both channels at once, and cost per acquired customer falls as the models compound.
On accounts where we run both on the same offline data, ROAS converges over the engagement. Consistent data quality produces that, not platform tactics. When the same signal reaches both algorithms, both optimize toward the same customer profile.
If Meta is optimizing on pixel events and you cannot tell which campaigns produce paying customers, the attribution gap is the problem to solve before any other campaign decision.
When the conversion is an offline revenue event rather than a pixel-fired page event, structure matters more. Each choice shapes the data the algorithm receives and how it weights that data.
Fragmented accounts split conversion volume across too many campaigns. Consolidating concentrates verified events into fewer learning pools, so each model reaches accuracy faster and there is less surface area for pixel/CAPI deduplication errors.
Creative is judged against the same KPI targets as the campaign. Decisions come from the downstream revenue data in Hyros, which ties revenue back to the specific ad, so each variant gets a verified revenue contribution, not a platform estimate.
Retargeting audiences are built from CRM segments, not pixel-based visitor lists. They contain people already qualified as prospects or customers, refreshed on a set schedule as new leads are added.
Rotation schedules are set from Meta frequency data, with new variants introduced before engagement declines. On unified-data accounts, rotation accounts for downstream revenue, not engagement alone.
The pixel fires from the browser when a user completes an action on a landing page. The Conversions API sends events directly from the server, including offline revenue that happens after the session ends. When the sale closes over the phone, in a CRM workflow, or by signed contract, the Conversions API is the only way to give Meta an accurate revenue signal.
We manage paid social accounts spending $100,000 or more per month, for businesses with a market big enough to grow 5x and the ambition to capture it.
When both platforms apply their own attribution to the same customers, each claims credit for conversions its ads touched within its window, so combined reported revenue overstates actual collected revenue. Hyros records the full sequence across both platforms and assigns credit by the verified customer journey, not each platform's independent window.
Any CRM that stores individual transaction revenue against an identifiable lead record and exposes it by API or export. Salesforce, HubSpot, GoHighLevel, and Zoho connect through native Hyros integrations; Stripe, Authorize.net, and most major processors connect by API or webhook. The event pushed via CAPI carries the original Meta click ID captured at lead entry, so Meta can match the revenue back to the ad interaction.
Standard lookalikes are seeded from pixel-based visitor or engagement lists, which include everyone who visited regardless of whether they paid. With offline data, lookalikes are seeded from CRM segments filtered to verified high-value customers, those who paid above a set threshold, or converted from specific campaign types. The result reflects your actual revenue-generating customers.
Meta begins updating its model as soon as verified events flow through the Conversions API. High-volume accounts see shifts within the first 30 days; lower-volume accounts accumulate the required data over 60 to 90 days before the model reaches accuracy that produces measurable cost-per-customer improvement.
View-through is excluded from the revenue model we use for bidding on every account. Meta's view-through window credits a conversion when a user merely saw an impression within a day of converting, without any click. Including it inflates Meta's reported contribution without confirming influence. Only conversions with a confirmed Meta click in the Hyros history determine Meta's budget allocation.
Meta's inventory covers Facebook, Instagram, Messenger, and the Audience Network. Placement decisions come from verified attribution data. Take Instagram as an example: where the Hyros history confirms it generates revenue-attributed interactions within the KPI targets, it stays in the mix with proportional budget. Placements that generate clicks without downstream revenue are excluded.
Next Step
Every proposal begins with an attribution audit: your current Meta tracking, CRM data structure, and the variance between Meta-reported conversions and CRM-recorded revenue. It produces a specific finding on what accuracy your stack can reach, before any engagement terms are discussed.
Request a Discovery Call →