Case Study

Vicious Biscuit

How proprietary transaction attribution connected digital advertising to real-world restaurant purchases across 11 locations—revealing which campaigns actually drive new customers.

Industry

Multi-location F&B

Locations

11

Solutions

Transaction attribution, paid media

Platform

CustomerMatch

The Challenge

11 Locations. Digital Ads Running. No Way to Know If They Work.

Vicious Biscuit is a fast-growing fast-casual restaurant brand with 11 locations across the Southeast. Like most multi-location restaurant brands, they invest heavily in digital advertising to drive awareness and foot traffic.

The problem: digital advertising platforms report impressions, clicks, and estimated conversions—but none of that connects to what actually matters. Did the person who saw the ad walk into the restaurant? Did they spend money? Were they a new customer or someone who was already coming?

Meta reports reach. The POS system records sales. Nothing connects the two. Most brands accept this blind spot as an unavoidable cost of doing business.

"We knew our ads were driving awareness. What we couldn't prove was whether they were driving revenue."

The Measurement Gap

Ad Platforms

Impressions, clicks, estimated conversions

Partial Data
BLIND SPOT

POS System

Transactions, revenue, items sold

Isolated Data

No connection between digital touchpoints and physical purchases

The Solution

The System Mediaura Built

A five-layer pipeline that ingests point-of-sale data, resolves customer identity, activates signals to advertising platforms, measures impact with statistical rigor, and delivers intelligence through AI.

Layer 01

Data Ingestion

Toast POS transaction data and Thanx loyalty program data from all 11 locations, normalized and ingested into a unified data pipeline on a daily cadence.

Toast POSThanx Loyalty11 LocationsDaily Ingestion
DeduplicationNormalizationUnified Customer DBCross-Source Matching
Layer 02

Identity Resolution

Customer records from multiple sources are deduplicated, normalized, and merged into a unified customer database—creating a single source of truth across all 11 locations.

Layer 03

Signal Activation

First-party customer data is activated through Conversions API (CAPI) to both Google and Meta, achieving a 9.3/10 match quality score—giving ad platforms real purchase signals instead of modeled estimates.

CAPI to GoogleCAPI to Meta9.3/10 Match ScoreFirst-Party Signals
Pearson CorrelationAdstock DecayLag AnalysisWeather Normalization
Layer 04

Measurement

Statistical modeling that measures the actual relationship between advertising spend and transaction volume—using Pearson correlation, adstock decay curves, lag analysis, and weather data normalization to isolate marketing's true impact.

Layer 05

Intelligence

A Claude-powered AI marketing analyst that can answer natural language questions about campaign performance, customer behavior, and revenue attribution—making the entire system accessible to non-technical stakeholders.

Claude AI AnalystNatural Language QueriesRevenue InsightsStakeholder Access
Why It Matters

Not a Dashboard. A Measurement System.

This is not another reporting layer on top of existing tools. It is a fundamentally different approach to marketing measurement—one that produces insights no dashboard can provide.

First-Party Data as Source of Truth

Instead of relying on platform-modeled conversions, we use actual POS transaction data and loyalty program records as the ground truth for measurement. The data comes from the cash register, not from a pixel.

New Customer Intelligence

Most restaurants can tell you total sales went up. We can tell you how many of those transactions came from customers who had never been seen before—the true measure of whether marketing is growing the business or just activating existing customers.

Statistical Measurement Independent of Platforms

Pearson correlation, adstock decay modeling, and lag analysis provide a measurement framework that does not depend on any advertising platform's self-reported metrics. The math works the same whether Google agrees or not.

High-Fidelity Signal Quality

A 9.3/10 match quality score on Meta means the platform's algorithms are receiving the cleanest possible purchase signals. This doesn't just improve measurement—it improves targeting, lookalike audiences, and campaign optimization.

Key Results

What the System Revealed

9.3/10

Meta Match Quality Score

First-party purchase signals matched at near-perfect fidelity to Meta's identity graph.

New

Customer Proxy Intelligence

For the first time, the brand can distinguish between new customer acquisition and repeat visit activation.

11

Store-Level Sensitivity Analysis

Marketing sensitivity measured independently for each location, revealing which stores respond most to advertising.

Independent

Measurement Framework

Statistical analysis that does not depend on any ad platform's self-reported conversion data.

Real-Time AI

Marketing Intelligence

Claude-powered AI analyst that transforms complex attribution data into plain-language answers. Stakeholders can ask questions like "Which campaigns drove the most new customers in Charleston last month?" and get an immediate, data-backed answer.

Want to Connect Your Advertising to Real Revenue?

If you're spending money on digital advertising and can't prove whether it drives actual purchases, you're making decisions based on incomplete data. We can fix that.

Transaction attribution isn't a dashboard—it's a measurement system built specifically for your business, your data, and your definition of success.