How to not be fired as a CMO in the next three years
In light of recent news of companies scrapping the CMO position, guest contributor Lee Arthur weighs in on what people must do to be successful in the role.
In light of recent news of companies scrapping the CMO position, guest contributor Lee Arthur weighs in on what people must do to be successful in the role.
While the title is extreme, the message is important: CMOs now more than ever are on the front line of delivering results. They have been tasked with taking what was an art and turning it into an informed science.
And with recent news of McDonald’s scrapping their CMO role (to be replaced by two SVPs), the position’s future — to some — seems up for discussion.
Now, more than ever, CMOs need to incorporate AI-based technology into their marketing efforts, forecasts, and ROI reports. Without that, a marketing team’s work is, at best, an informed guess. And at worst, it could be completely wrong.
Here are the four main takeaways to keep in mind for a successful future as a CMO:
Even if you understand probability and statistical analysis, you can no longer cope with the explosive growth in marketing choices for audience, data, and measurement tools.
The methods by which marketing teams measure their ROI on spend are typically rules-based (deterministic models), often provided by their vendors (who have their own incentives). Examples include continued use of first click and last click models from Google and Facebook.
People often say, “all models are wrong, some are useful.”
These rules-based models that don’t use machine learning to deliver probabilistic answers have the highest error rates.
Sometimes the “sum” reaches 150% or even 500%!
The good news is that for the price of just one member of your marketing team, you can add a machine learning intelligence system to your decision-making. Think of it as a new really smart team member, like the ‘Data’ character on Star Trek. This machine learning “team member” will give the rest of your team the ability to use probabilistic inference to make highly accurate predictions for the future.
Said simply, it will take in all your click, revenue, and cost data and show you the holes in your models.
With that information, you can then turn off what is not working and increase what is working. This level of detailed statistical analysis might take humans years to calculate.
That’s a lot of money left on the table.
Ps — we created a free attribution course with Facebook and Fospha where you can learn more about the potential for intelligent attribution.
Disclosure: Lee Arthur leads the US office of Blenheim Chalcot, a digital venture builder. Fospha and ClickZ are within their portfolio.