How My Ex-Boss Built Happy Coefficients with Excel Solver and Saved The Day

Enter the Solver Samurai

Imagine the open-plan office of a global media agency next to Sydney Harbour Bridge. The coffee machine is working overtime. PowerPoints are flying left and right.

And there he is — my ex-boss. A man who had conquered the worlds of media effectiveness and efficiency measurement for Australia and New Zealand.

Forget Python.

Forget Google Meridian and Google Lightweight.

Forget Meta Robyn.

Forget cloud-based AI dashboards with amazing charts.

His secret weapon?

👉 Excel

And his sidekick?

👉 Solver

If this were a Marvel movie, his superhero name would be The Solver Samurai armed only with pivot tables and a dangerous amount of Ctrl+C, Ctrl+V, and Ctrl+Z reflexes.

The Holy Grail — Happy Coefficients

In Marketing Mix Modelling, coefficients are life. They’re the numbers that tell you:

• Do TV Ads actually drive sales or just make the CMO feel important?
  
• Are Google Ads worth the budget, or just an expensive way to buy clicks from your own brand name?
  
• Do marketing 4ps lift revenue, or just teach customers to never buy full price again?
  

But coefficients are tricky little beasts.

Too high?

This is too good to be true.

Too low?

Are you saying media doesn’t work?

In other words:

• If coefficients look bad → the CMO volunteers you for a surprise budget review
  
• If coefficients look fake → the media agency makes you the star of PowerPoint purgatory
  
• If both look unhappy → your job is in danger, and no free coffee and muffins in the kitchen can save you
  

So my ex-boss created the Happy Coefficient Rule:

👉 Align share of spend (budget percentage) with share of impact (results percentage).

Balance achieved = happy coefficients

Happy coefficients = happy stakeholders

Happy stakeholders = all jobs remain safely at their desks

Excel Math Solver, the Drunk Intern

Let’s talk about Excel solver.

Solver is that intern you don’t fully trust but can’t live without. Sometimes brilliant, sometimes chaotic:

• It finds the right coefficient like a genius → ✨
  
• Or it gives you negative TV impacts → 🤡
  
• Or it crashes Excel with “Not Converged” → 💀
  

Running Excel solver is like texting your ex: you’re never quite sure what you’ll get back.

So what did my ex-boss do?

He ran Solver again. And again. And again.

TV ads? Solver until happy.

Google ads? Solver until happy.

Other media? Solver until happy.

Solver was basically a casino slot machine — keep pulling the lever until the numbers look good. 

The 108-Model Nightmare

Then came the FMCG monster project.

• 6 states in AU market
  
• 6 retail networks
  
• 3 product categories

That’s 108 models. Yes, one hundred and eight.

The boss divided the team into three squads. Each got 36 models. The instructions were crystal clear:

• Open Excel
  
• Build regression using Ordinary Least Squares (OLS)
• OLS sets up the regression, but let’s be honest — it’s just doing the background work. The real fun starts when we fire up solver and make the coefficients behave
  
• Set “happy” share-of-spend vs. share-of-impact targets
  
• Run solver until nausea
  
• Repeat until fingers lose sensation

Two weeks of this. No daylight. No joy. Just Excel solver.

At one point, the mere word “Excel” during lunch triggered gag reflexes. Imagine someone trying to enjoy a sandwich and another analyst whispers “solver…” → instant vomit.

It was like an Excel cult initiation. By week two, half the team considered switching careers to goat farming.

The Happy Ending

But against all odds, the team survived. Solver eventually spat out 108 sets of coefficients that passed the Happyness Test.

On presentation day:

• The PPT was slick
  
• The dashboards sparkled
  
• The coefficients looked magical
  

The results?

• The media agency: “Yes! Media works, look at those beautiful impacts!”
  
• The CMO: “See? My budget decisions are pure genius.”
  
• The CFO: “Fine, I’ll stop asking about cutting TV spend.”
  
• My ex-boss: “Solver, you magnificent drunk intern, you did it again.”
  

Everyone smiled. Everyone clapped. Everyone kept their jobs.

Even Excel felt proud, humming quietly in the background: “Not Responding… just kidding, I’m fine.”

Meme-worthy Truth Bombs

To make sure we all remember this legend:

• Excel Solver is like Thanos: “I am inevitable.”
  
• Every analyst knows the true horror movie jump scare isn’t Jason… it’s #VALUE! at 2am.
  
• Linear regression: the most straightforward tool that somehow causes the most existential crises.
  
• A “happy coefficient” is like a corporate unicorn: half math, half politics, 100% survival.
  
• MMM in Excel? More like “Marketers Must Meditate” before running solver again. 
   

  Lessons Behind the Laughter

Okay, let’s get serious (just for a sec).

Linear Regression (OLS) is a natural and powerful way to model marketing and media effects — but on its own, it’s like showing up to a party in slippers. Sure, you’re there, but nobody’s impressed.

Enter Excel solver, the over-caffeinated sidekick who actually gets things done. It wrestles the coefficients into place, tames the rogue numbers, and basically turns your “meh” OLS into a superhero-level marketing insight.

Coefficients aren’t just boring numbers — they’re like tiny little diplomats. They negotiate trust between client and agency, justify budgets, and keep partnerships from spontaneously combusting. Ignore them at your peril.

Marketing Mix Modelling is political — a model isn’t just about math, it’s about believability. You can have perfect accuracy, but if your CMO raises an eyebrow, you might as well be predicting unicorn sightings.

At the end of the day, stakeholder happiness and client relationships matter just as much as mathematical perfection. After all, a perfect model that nobody trusts is like a perfectly baked cake that no one’s allowed to eat — tragic, useless, and kind of sad.

Final Curtain Call

So next time you curse at Excel freezing, remember my ex-boss — the Solver Samurai.

The man who built 108 models, ran solver like a mad scientist, balanced coefficients like a diplomat, and somehow kept everyone smiling.

Marketing mix modelling is part science, part politics, and part circus act. But when done right, it’s not just numbers — it’s job security, client love, and agency survival.

All thanks to one tool: Excel solver, the drunk intern who saves the day.