Marty Smithson
Marty Smithson

Marty Smithson

I've directed $25M+ in ad spend, and I decide where the next dollar goes with MMM, geo and incrementality tests, and the weekly cash MER leadership reads.

I run BI and performance marketing at Black Clover, a golf apparel brand in Lehi, and I report to the CEO. Day to day that means building the models and tests that tell us what's working, then making the call on where the money goes.

Head of Business Intelligence & Performance Marketing, Black Clover

Founder and Data Analytics Consultant, McFly Ads

BS, Utah Valley University

martysmithson04@gmail.comLinkedInResume PDFMcfly Analytics Shopify App

Approach

How I work

  1. I've had to spend the money

    I've moved real budget on Google, Meta, TikTok, Amazon, and Microsoft. When I read a model, I read it knowing I'm the one who has to act on it.

    The Black Clover case →

  2. I build my own tools

    I built a Shopify app that shows merchants what their customers are actually worth. It's live on the App Store, and I wrote up why I built it the way I did.

    Live app (sample shop) ↗Read the explainer →

  3. Cash comes first

    Ad platforms grade their own homework. I check every one of them against what the business actually booked, divided by everything it spent on ads. We call that cash MER, and it's what sets the budget. Platform ROAS stays on the page so we can see what each one claims.

    MER dashboard demo →How I reconcile platform numbers with cash →

  4. Less Google, better return

    Google was getting about half our paid budget. I took it down to 35%. We spent more overall and total ROAS still went up. I put five years of history on one dashboard so the CEO could check the whole picture.

    Read the case →

Work

Cases

Black Clover · Paid mix

Moving budget off Google without losing return

Head of Business Intelligence & Performance Marketing · Black Clover · Feb 2025 – present

I report to the CEO and own where the paid budget goes. Meridian shows me where each channel stops paying back. When two channels both claim the same sale, a GeoLift test or an RCT settles it.

  1. The question

    Google, Meta, and Microsoft all count their own conversions, and some of those are the same orders. So which one gets the next dollar?

  2. The call

    Google had about half the paid budget. I cut it to 35% and moved the difference into the other channels.

  3. What happened

    We spent more in total and ROAS still went up. More spend usually drags return down, so that was the part I cared about.

  4. What the CEO saw

    I built a dashboard with five full years of history so the CEO could see what changed without me choosing the dates.

Read the full case

Black Clover · Executive BI

Getting spend and sales into one place

Head of Business Intelligence & Performance Marketing · Black Clover · Feb 2025 – present

The measurement only helps if leadership is looking at the same number. I built what sits under it: pipelines from the ad platforms into NetSuite, a move off Domo that took a ~$20K monthly bill to under $30, real-time MER dashboards, and the executive reporting.

Read the full case

Nutricost

What each of four ad platforms actually drove

Advertising Data Scientist · Nutricost · Aug 2024 – Feb 2025

I built the measurement for a paid portfolio running about $2M a month across Amazon, Google, Meta, and TikTok. The biggest finding: Amazon Ads was taking credit for sales organic search would have brought in anyway.

Read the full case

McFly Ads

My consulting practice, and the Shopify app I built

Founder and Data Analytics Consultant · McFly Ads · Apr 2020 – present

McFly Ads is my consulting practice. I work with e-commerce and DTC brands on measurement strategy, attribution, and media optimization. Separately, I built and launched the Mcfly Analytics Shopify App, which shows merchants cohorts, lifetime value, and five years of sales. It's live on the Shopify App Store.

Read the full caseLive app (sample shop)mcflyanalytics.com

Measurement

Five demos you can try

Five small tools I built so you can poke at how I think. The mix model is open below.

  1. Demo · synthetic dataMMM budget reallocationWhere should the next dollar go?A year of weekly sales split by channel. Move budget between four channels and watch the whole year redraw, along with what the next dollar buys in each one.
  2. Demo · synthetic dataGeo test readoutDid the spend cause the sales?Test markets against a matched control. You get the lift, how sure we can be about it, and a call: keep it, cut it, or keep testing.
  3. Demo · synthetic dataExperiment power calculatorHow big, and how long?How big a test has to be, and how long it has to run, before the answer means anything.
  4. Demo · synthetic dataAttribution comparisonWho gets credit, and does it add up?The same customers credited three ways: last click, data-driven (Shapley), and what was actually incremental. Then every channel's claims get checked against the orders.
  5. Demo · synthetic dataExecutive MER dashboardWhat did we book against what we spent?The weekly number I'd put in front of a CEO: what we booked against what we spent, next to what the platforms say they drove. One week is blank on purpose because its data isn't in yet.

Method write-ups

  • From MTA to macro-measurementWhy user-level click paths stopped being a reliable budget source, dated and sourced, and what replaced them: modeled conversions, MMM, lift tests, and MER.
  • MMM workflowI start from the budget question and build the dataset from booked sales. Lift tests anchor the model, and what leadership gets is a recommendation with a range around it.
  • Designing a geo testDecide what each result would mean before launch. Then pick markets that move together, size the test from their pre-period noise, and judge the whole interval against break-even.
  • Reconciling platform vs cashPick one source of truth for sales and spend and explain the gap line by line. Then de-duplicate the claims and lead with cash MER. Platform ROAS stays on the page as context.
Demo · synthetic data

MMM budget reallocation

A year of weekly sales for a made-up brand with four paid channels. Move the sliders or hit rebalance and the whole year redraws.

Predicted weekly revenue

$588k

0.0% vs. starting plan

Weekly spend

$155k

Starting plan $155k

Blended MER

3.80×

Revenue over total spend

Next dollar to

Paid social

From Retail media

Bars show

52-week totals: sales $30.69M, spend $8.06M, ROAS 1.86×

Revenue per weekROAS$0$200k$400k$600k$800k0.0×2.0×4.0×6.0×W1W6W11W16W21W26W31W36W41W46W51
Baseline (no paid media)Paid searchPaid socialOnline videoRetail mediaSales (left axis)Sales at starting plan (left axis)ROAS (paid revenue ÷ spend) (right axis)

Paid search

Next dollar 1.19×

$45k
Revenue
$96k
Average ROAS
2.14×

Paid social

Next dollar 1.71×

$60k
Revenue
$114k
Average ROAS
1.90×

Online video

Next dollar 1.44×

$20k
Revenue
$16k
Average ROAS
0.82×

Retail media

Next dollar 1.18×

$30k
Revenue
$61k
Average ROAS
2.05×

Sliders set the average week. Each channel's spend moves around that average with a seeded seasonal pattern (video runs in flights, retail media spikes around two sale events), carries over into later weeks, and runs through a Hill saturation curve. The small curve in each tile is that channel's response; the dot is where you are on it. Baseline sales without paid media average $300k a week. The parameters are made up and stand in for a fitted model's posterior means.

Synthetic numbers generated in your browser. Not client or employer data.

How to read this demoHow I'd approach an MMMThe shift from MTA to macro-measurement

How I set budgets

Platform ROAS, lift, and cash are different numbers

A platform's ROAS is that platform's claim. A lift test tells me what was incremental. Cash is what the business booked against what it spent. I set budgets on the last two.

Contact

Get in touch

martysmithson04@gmail.comLinkedInResume PDFMcfly Analytics Shopify App