Growth & performance marketing · Los Angeles

Tony Lee

Athlete’s discipline. Analyst’s rigor. A writer’s need to explain it plainly. This is the work, and the thinking behind it.

Cover · Experience

The résumé version.

  • Shopify, Yamibuy, StarzIn-house experience, alongside agency roles at Monks, OMG23 and Reprise
  • 13+ yearsWorking in growth and performance marketing
  • Social Skills, Season 3Featured in Meta for Business’s series on how brands find customers
  • Across the growth teamAcquisition, measurement, budget decisions and people

That’s the short version. The decisions take a little more explaining.

Mantle · Decisions

Why the work looked that way.

  • Friction by DesignAn extra step before a Shopify landing page, and the highest session-to-lead conversion rate on record
  • Would they have bought anyway?The question behind a different approach to acquisition at Yamibuy
  • 70 / 20 / 10How I divide a budget between proven work, promising ideas and new experiments
  • Building with AIWhat went into this site, and what remains my responsibility

The results fit on a résumé. The reasons need a page each.

Core · The person

The person doing the work.

  • Division I, then the newsroomPlayed golf at St. Bonaventure, then gave it up to found and run The Intrepid, an online college paper
  • Responsible for the teamPeople should feel secure enough to ask questions, try things and tell me when I’m wrong
  • Henry’s dad. Eun Bi’s husband.Learning as we go
  • Numbers person. Full effort.228 rounds logged, every shot tagged. I measure my game like a media plan

Selected work

Three places to start.

Three decisions from the work, with the context behind the results.

  1. 01 Friction by DesignAdded an interactive step before a Shopify landing page to qualify incoming traffic. Shopify RECORD Highest session-to-lead conversion rate on record at Shopify
  2. 02 Would they have bought anyway?Reconsidered where acquisition spend was going. Yamibuy −15% CAC vs. baseline · volume up 25% YoY
  3. 03 Zero to twelveJoined as the first U.S. employee and built a book of business. U.S. launch 0→12 Accounts in eighteen months

Read the work

Experience

People and projects.

Brands I’ve worked with, conversations on camera, and recommendations from people who know the work.

Hover or tap a flag for the role or agency.

  1. DECLÉOR5WPR client
  2. SoffeThe Variable client
  3. Party CityZimmerman Advertising client
  4. Sam's ClubZimmerman Advertising client
  5. sweetgreenOperam client
  6. NintendoReprise Media client
  7. 20th Century FoxOperam client
  8. StarzIn-house
  9. FX NetworksOMG23 client
  10. ShopifyIn-house
  11. FandangoConsulting
  12. SearchlightConsulting
  13. A24Consulting
  14. GoogleMonks client
He impressed me with his ability to hit the ground running as a leader… he truly cared about their development.
Andriana AngelevskaReported to Tony · OMG23
He is an incredible leader and has the ability to motivate those around him.
Danielle WhitingReported to Tony · Growth Director
Working under Tony in the past was amazing for my career development.
Allan Banker IIReported to Tony · Paid Media Lead

On camera · Social Skills, Season 3 · Meta for Business

More experience and recommendations

70 / 20 / 10

How I divide a growth budget.

Proven work pays off often and modestly. Experiments rarely pay off, but when one does it changes what the proven budget should fund next. My split is 70% proven, 20% promising, 10% new experiments. Move the sliders and watch the expected value change.

Enter a quarterly budget and move the sliders to see the dollars in each category. The shares always total 100%.

70%

Work with evidence of commercial return, checked against what would happen without the spend.

20%

Ideas with a reason to believe, and a question still to answer.

10%

Tests with a clear question and a limit on what we’re prepared to spend to answer it.

Proven work Promising ideas New experiments Each dot is 2.5% of the budget
Proven work
$420,000
Promising ideas
$120,000
New experiments
$60,000

Expected value, this quarterIllustrative

BucketPays offIf it doesExpected back
Proven work
Promising ideas
New experiments
Portfolio, per $1 spent

On this quarter’s math alone, going all-in on proven work wins. The 20 and the 10 pay for next year: proven channels saturate, and the only way to find the next one is to fund the search.

That’s my split: proven work carries the quarter, and new ideas still get a real chance.

Chance and payoff figures are illustrative assumptions, not results from any client or employer. Not a forecast or a guarantee of return.

Why this split

Did this work?

What happened because of the ads?

Here’s the math I’d walk a finance partner through. Pause ads in a few markets, keep them running everywhere else, and compare the gap against how those same markets behaved before the test. Then ask how sure you can be. Pick markets, run six simulated weeks, and every step of the math appears below the result.

Alabama Alaska Arizona Colorado Florida Georgia Indiana Kansas Maine Massachusetts Minnesota New Jersey North Carolina North Dakota Oklahoma Pennsylvania South Dakota Texas Wyoming Connecticut Missouri West Virginia Illinois New Mexico Arkansas California Delaware District of Columbia Hawaii Iowa Kentucky Maryland Michigan Mississippi Montana New Hampshire New York Ohio Oregon Tennessee Utah Virginia Washington Wisconsin Nebraska South Carolina Idaho Nevada Vermont Louisiana Rhode Island
  • Ads continue
  • Ads paused
  • Markets still running ads

1 · Choose holdout markets

Ads pause in the markets you select and continue everywhere else.

No markets selected

Choose from a list

2 · Run the simulation

3 · Read the result

Choose markets, then run the simulation.

Markets and sales are simulated. The method is a geo holdout read with a difference-in-differences ratio; the 95% interval is ±1.96 standard errors of the six weekly estimates. It illustrates the logic; it doesn’t design or validate a real experiment.

How I think about measurement