Episodes

Andrew Grosso: What. So What. Now What.

Andrew Grosso, Co-Founder and CPO of Pickaxe, on why a drama major became obsessed with data, how marketing mix modelling helps broadcasters and streamers allocate spend across awareness and performance channels, and why the three-question framework of what, so what, and now what turns data into executive action.

Apple Podcasts Spotify
 ·  The Business of Marketing  · S2 E23  · 43 min
Read the interview

"Search always looks like it's going to bring all your customers in because it's the last click. But how much better does search get when you put up a billboard?"

The Conversation with Justin Cooke

Why data is useless without a story to explain it, and how the what-so what-now what framework turns analysis into executive action

Andrew Grosso studied drama, moved to New York, answered a job ad at a hedge fund that owned an AOL competitor, and became fascinated by the work of a data modeller who could predict exactly how many hours of call centre capacity a CD mailout campaign would require weeks in advance. That fascination led him through data analytics, marketing science, and eventually to co-founding Pickaxe, a platform that provides AI-automated marketing insights and marketing mix modelling to broadcasters, streamers, and direct-to-consumer brands.

In this conversation Grosso makes the case that the analytical training of a drama student and the discipline of a data scientist are solving the same problem: given a complex story with a lot of moving parts, how do you present it in a way that keeps your audience engaged and tells them what to do next? He applies this framework directly to how marketers should present data to executives: not as a report, but as a three-act structure in which what happened gives way to why it matters, and why it matters gives way to what to do now.

Key Takeaways

  • Data is useless without a story to explain what it means. The analytical skills and the storytelling skills are solving the same problem.
  • What. So what. Now what. That is the structure every data presentation to an executive should follow. Stop at the what and you have produced a report, not a recommendation.
  • Search takes all the credit in last-click attribution. Put up a billboard and watch organic search increase. Marketing mix modelling captures that relationship. Last-click does not.
  • AI-automated insights make marketing mix modelling accessible to teams that could not previously afford the cost or time of traditional econometric approaches.
  • The same problem exists for broadcasters and streamers as for any brand: how do I allocate my spend to hit my subscriber and revenue goals most efficiently?

Topics Covered

  1. 01 Why data is useless without a narrative to explain its meaning
  2. 02 The what-so what-now what framework for presenting data to executives
  3. 03 Marketing mix modelling for broadcasters and streamers: how to allocate spend across awareness and performance
  4. 04 Why last-click attribution flatters search and undercredits upper funnel investment
  5. 05 AI-automated marketing insights: making econometric modelling accessible to more marketing teams

Key Exchanges

05
01 Tell me about Pickaxe and what you do.

At Pickaxe we do two things. We work with a number of companies, mostly large media and entertainment, lots of broadcasters in the UK like BBC and ITV and Sky, and also in the US from Peacock to Fox News and MSNBC at the same time. We said, clearly we are doing something. We also work with direct-to-consumer companies like Hallmark. We have a software platform with tools involving AI automated insights and a mix product designed just for marketers. And we help with data challenges.

Grosso's description of working with Fox News and MSNBC simultaneously is a deliberate illustration of Pickaxe's analytical neutrality. The platform is measuring the commercial effectiveness of marketing decisions, which is not a political judgment but a mathematical one. The fact that ideologically opposed media organisations trust the same platform for that measurement is evidence of the genuine utility of what Pickaxe does.

02 How did a drama major end up building a data analytics company?

At its core, the data is useless without some sort of story to explain what it means. You are always thinking about the data and trying to explain it to the EVP of marketing who is very busy and just wants to know: so what? You have told me what about this data, but so what? What is important about what you are saying? And then what is the now what? What do I do next? And I think that is the dramatic training.

Grosso draws a direct line between the skills he developed as a drama student and the skills required to present data analytically. Both require understanding an audience, structuring information so it builds toward a conclusion, and delivering something that changes what the audience believes or intends to do. The three-act structure of what, so what, now what is a dramaturgical framework applied to data analysis, and it is one of the clearest articulations of why data literacy and communication skills are not separate disciplines but complementary ones.

03 How does marketing mix modelling work for broadcasters and streamers?

Marketers want to know how much they need to spend to hit their goals of subscribers or purchasers or customers, and then where to spend. I need ten million dollars. How do I allocate that most efficiently? For them it is the same challenge. It is either I want to know how many purchases I am going to get, or how many unique viewers I am going to get so that I can get ad dollars.

The fundamental question that marketing mix modelling answers for broadcasters and streamers is the same one it answers for any brand: given my budget and my commercial goals, what is the optimal allocation across channels? The additional complexity for media companies is that they are simultaneously managing subscription acquisition and advertising revenue, both of which respond differently to different marketing activities and require separate models that need to be reconciled in a single allocation decision.

04 Why does last-click attribution systematically mislead marketers about the value of upper funnel activity?

Search always looks like it is going to bring all your customers in because it is the last click and it is getting this credit. But how much better does search get when you put up a billboard? Can you tell that impact? That is a thing for marketers that carries on this tradition of, hey, we do not have all of the data points, but we can do some math and give you a range of expectations.

The billboard and search example illustrates the core problem with last-click attribution. When a brand runs an out-of-home campaign, brand awareness increases, which drives people to search for the brand, which shows up as organic or paid search conversions. Last-click attribution gives the credit entirely to search. The billboard gets nothing. This systematically undervalues brand and upper funnel investment and leads marketers to over-invest in lower funnel channels at the expense of the demand-generating activity that makes those channels work.

05 What role does storytelling play in making data actionable?

There was this great article in The Times about Nick Cohn explaining the difference between polling techniques. And he is such a great writer that he could explain why recentring of polls is a bad technique in Florida because the electorate has shifted. For marketing, the same principles apply. You are always thinking about the data and trying to explain it to the EVP of marketing who is very busy and does not have a lot of bandwidth. Just wants to know: so what?

Grosso uses the election polling analogy to make a subtle point about the relationship between data and narrative. The raw numbers from a poll, or from a marketing analytics platform, mean nothing without the context to interpret them. Is this result good or bad? Is this trend normal or anomalous? What should we do about it? Those questions require judgment, context, and communication skills. The analyst who can answer all three questions, not just produce the numbers, is the one who creates commercial value.

S2 E23Season & Episode
43 minDuration
3 Questions every data presentation must answer: what, so what, now what
100% Drama major who became obsessed with data modelling

"You need to keep your audienceand then give them something and put it into context."

Hear Andrew on
The Business of Marketing
Season 2 Episode 23 43 min
Apple Podcasts→ Spotify→ Andrew's full profile→ Andrew on LinkedIn→
Read the full transcript
Lightly edited for readability.

Justin Cooke Tell me a little bit about Pickaxe.

Andrew Grosso At Pickaxe we do two things. We work with a number of companies, mostly large media and entertainment, lots of broadcasters in the UK like BBC and ITV and Sky, and in the US from Peacock. We also work with direct-to-consumer companies and retailers like Hallmark. We have a software platform with two tools involving AI automated insights and a mix product designed just for marketers. And we help with data challenges, from strategising how to start a new business to data engineering and data science.

Justin Cooke How did a drama major end up doing what you are doing now?

Andrew Grosso I moved to New York and realised it was very expensive. So while I was directing theatre, I answered a job ad and ended up working at D.E. Shaw, which owned Juno Online, an AOL competitor. They were hiring people straight out of university. I got very interested in customer service reporting and creating reports for marketing. From there I ended up doing product management for internal tools and data tagging implementations and started working into data analytics, which turned into data science and Pickaxe.

Justin Cooke What is the link between dramatic training and data work?

Andrew Grosso At its core, the data is useless without some sort of story to explain what it means. For marketing, you are always thinking about the data and trying to explain it to the EVP of marketing who is very busy and just wants to know: so what? You have told me what about this data, but so what? What is important about what you are saying? And then what is the now what? What do I do next? And I think that is the dramatic training. You need to keep your audience and then give them something and put it into context.

Justin Cooke Give me the link to Pickaxe operating inside the world of broadcast and streaming.

Andrew Grosso Marketers want to know how much they need to spend to hit their goals of subscribers or purchasers, and then where to spend. I need ten million dollars. How do I allocate that most efficiently? For them it is either how many purchases I am going to get, or how many unique viewers I am going to get to attract ad dollars. And you need to figure out how to take your budget on Google and how much do you spend on awareness, how much on YouTube versus search. Search always looks like it is going to bring all your customers in because it is the last click. But how much better does search get when you put up a billboard?

Keep Reading