Like what you see?

Book a Free Consult →
Accepting new engagements

Your ad platforms are lying
about what’s driving revenue.

Meta says it drove the sale. Google says the same. Klaviyo takes credit too. Add it all up and you’d need 10x your actual revenue to make the numbers work. I build custom statistical models that cut through the noise — isolating real channel impact from platforms taking credit for sales they didn’t drive.

Campaign-level MMM. Under 2 weeks. No subscription. You own everything.

Campaign Level
Not Just Channel Level
14 Channels
Unified in One Model
< 2 Weeks
To Delivery
No Subscription
Yours Permanently
Pythonscikit-learnstatsmodelsPyMCRidge RegressionBayesianHosted on Cloudflare$0/mo forever
The Problem

Your ad platforms are not telling you the truth.

Platform ROAS Double-Counting

Add up your platform ROAS and you’d need 10x overall returns. Meta, Google, and TikTok all take credit for the same conversion. Last-click assigns 100% to whoever touched the customer last — even if six other channels did the work.

SaaS Attribution Doesn’t Solve It

Attribution platforms cost $1,500–$18,000/year for template models, black box outputs, and a dashboard you’ll never own. The fundamental measurement problem — separating true incremental lift from noise — remains unsolved.

Email’s Real Impact Is Invisible

Klaviyo claims every sale within 5 days. The model tells the truth. Email revenue lags 3 to 14 days. Without modeling that delay, you’re either firing your best channel or crediting it for sales it didn’t drive.

Live Demos

This is what we build for clients.

Not a mockup. An actual MMM model running in your browser. Adjust the sliders — watch the math update in real time.

📊 Marketing Mix Model — Budget Optimizer Live Interactive

Reallocate $300K Across 5 Channels

Predicted Revenue by Channel
Marginal ROAS — Diminishing Returns

🔍 Platform Claims vs. MMM Truth Live Interactive

See how platforms over-claim revenue — and what the model actually attributes

Methodology

Campaign-level precision. Not channel-level guesses.

Statistical Models, Not Pixels

Regression models trained on your actual revenue history — not platform tracking pixels that break with every iOS update. Aggregate data. Privacy-safe. No cookies.

14 Channels, One Truth

Google Search, Shopping, Meta, TikTok, Amazon, Microsoft, Pinterest, email, SMS, influencer, organic, direct, affiliate, offline — unified in a single model with marginal ROAS for each.

Adstock & Lag Built In

We model how long each channel’s impact decays after spend stops, and the delay between email send and revenue response. Most tools assume instant impact.

Services

What gets built.

🔬 Marketing Mix Modeling & Attribution

The core offering. Custom regression MMM with campaign-level granularity, diminishing returns curves, email lag modeling, adstock decay, and interactive budget optimization. Calibrated to your business, delivered in under 2 weeks, with full code handoff. Unlike SaaS tools, you own everything and pay nothing after delivery.

Ridge RegressionBayesianIncrementalityLift TestsCampaign Level14 Channels

Analytics Frameworks

Event taxonomy, conversion tracking, cross-platform data unification, and KPI definition.

GA4SegmentMixpanel

Dashboards & Reporting

Real-time KPI dashboards, MER tracking, executive reporting, automated alerts.

DomoTableauLooker Studio

Data Engineering

ETL pipelines, data warehouse design, cross-platform integrations, quality checks.

SQLBigQuerySnowflake

Channel Optimization

Full-funnel paid media across Google, Meta, TikTok, Amazon. ROAS optimization.

GoogleMetaTikTok
Results

Selected work.

E-Commerce • Apparel

Cross-Channel Measurement Framework & Budget Optimization

Unified Google, Meta, and Microsoft data with ERP via automated ETL pipelines. Model revealed Google Search was past efficient marginal returns while Meta had significant budget headroom. Reallocated spend accordingly, improving blended MER while cutting waste.

22%
Over-Spend Identified
$3.6M+
Annual Budget Measured
Real-Time
MER Dashboard
Health & Wellness • DTC

Multi-Platform Attribution & Incrementality Analysis

Built measurement framework for 7-figure monthly portfolio across Amazon, Google, Meta, and TikTok. Found Amazon Ads cannibalizing organic at net incrementality far below platform claims. TikTok upper-funnel lift appeared in branded search 7–10 days later — invisible to last-click.

2.1x
True vs. 4.8x Claimed
$2M/mo
Spend Measured
4
Platforms Unified
Pricing

Pay once. No subscription. Yours permanently.

No vendor lock-in. No annual contracts. Full code handoff on every engagement.

Quick Win
MER Diagnostic
$750
You provide the data. Delivered in 48 hours.
  • Full attribution gap analysis
  • Channel-level MER breakdown
  • Spend leakage identification
  • Written recommendations
  • 45-min walkthrough call
Get Started →
Ongoing
Measurement System + Support
$10,000–$15,000
First year. Then $3K–$5K/year for quarterly reruns.
  • Everything in MMM Sprint
  • 4 quarterly model reruns included
  • Quarterly strategy call
  • Dashboard updates as channels change
  • Priority support
  • Year 2+: $3K–$5K for reruns only
Get Started →
Triple WhaleNorthbeamMcFly Ads
Year 1 cost$2,400–$6,000$18,000+$5,000–$15,000 one-time
Year 2 cost$2,400–$6,000$18,000+$0 or $3K–$5K reruns
3-year total$7,200–$18,000$54,000+$5,000–$25,000
You own the codeNoNoYes
Custom to your dataTemplateTemplateBuilt for you
MMM with diminishing returns per campaignNoChannel-levelCampaign-level
Email lag modelingNoNoYes
Vendor lock-inYes (subscription)Yes (subscription)None
Cumulative Cost: SaaS Subscriptions Keep Climbing. McFly Ads Stays Flat.
Let’s Talk Measurement

Tell me about your business.

Drop your info. I’ll reply within 24 hours with initial thoughts.

✓ Under 24hr response ✓ $750 MER Diagnostic available Full form with details →
Marty Smithson
Built by Marty Smithson. 10+ years in marketing analytics. Managed multi-million dollar ad portfolios across Google, Meta, TikTok, Amazon, and Microsoft. Every model I build comes with an interactive dashboard, plain-language documentation, and full code handoff. LinkedIn · Learn more Learn more →#x2192;