Creative optimization: comprehensive approach and tools
In this article, I challenged myself to help you understand what a CMP is. Why? In our industry, digital marketing, it has become practically impossible to talk about creative without talking about audiences. We want to understand how to break them down, how to develop strategies around them. If you want to understand what a CMP is, you have to understand audiences. And everyone should know about CMPs, not just media buyers.
Adviso’s optimization processes are extremely rigorous. That’s to be expected, when you consider the scientific approach we take:
The secret to successful optimization is in the frequency and uniqueness of the variable you are trying to isolate. It is therefore recommended that you make one big change per week, rather than dealing with multiple tiny details.
What can you change to improve a campaign’s performance? There are lots of variables you can test, but regardless of which you choose it’s important to ensure your tests follow a logical structure.
In any campaign, audiences are generally broken into multiple segments. To get good results, the creative needs to be different for each of the six levers. We are therefore talking about a significant quantity of creative. You then have to prepare a series of weekly hypotheses for each element. Essentially, if you follow the diagram above, you’ll see that it will roughly take 18 weeks to optimize your creative.
Do you know the impact creative has on a campaign? According to Google Media Lab Research 2016, the answer is 70%. It is huge!
The value of creative to campaign performance is such that we should never allow ourselves to put off its optimization. It’s a paradigm that requires reflection. How can we put creative at the center of our optimizations?
Programmatic came onto the scene with an ambitious promise: the right message, to the right person, at the right time. In reality though, programmatic just speaks to the right person at the right time. The message doesn’t really play in. That might explain why so many companies invest so much into this search for solutions; personalization is like a gold rush!
Have you heard of Avinash? Flashback to 2008, and the rise of the famous See-Think-Do model, also known as Top-Mid-Low.
For example, right now, you are probably trying to understand what a CMP is. Let’s agree that you are closer to a conversion point than if you were scrolling through your news feed in a meeting that was running long. Using this model from 2008, the goal was always to drive back to the ecosystem, and a final binary ending: conversion or no conversion. It was a simpler time… Budgets were primarily invested in the bottom of the funnel, as close to the point of conversion as possible, then the middle, then finally, if there was budget left over, the top. Brand image was also far less important in digital because it simply made more sense to put your money on measurable conversions.
Then, the notion of audience was added to the model thanks to the rise of behavioural targeting with cookies. Visitors, who were no longer necessarily unknown, could be retargeted. A nice advancement but the problem remained – two-step conversions. The highly unsubtle conversion messaging remained the same: BUY. NOW.
Faced with this challenge, our research and development team created a model that allows for even greater precision, the AMFM model. No, it has nothing to do with the radio. The acronym stands for Audience Micro Funnel Migration. The intention behind the model is to stop looking at the platform and instead go where the consumer is in relation to you, therefore getting a better understanding of who they are and what path will lead them to a conversion.
AMFM Model developed by our Data science and technology department
But have we successfully implemented the personalization we were talking about earlier? Not yet.
To present the right message, it’s important to take into account both AMFM segments, and personalization signals. These signals can be placed in three categories:
- Behaviour and interests
- Media points of contact
The data has three possible points of origin.
This data could include age, sex, habits, interests, income, etc. Concretely, this data would be used to present one creative to group A (men, for example), another creative to group B (women, say) and a third, neutral variable. To validate whether the creative has the right messaging, the creative that’s been personalized for the specific groups would have to perform better than the neutral creative.
A problem we see often is that creative is attributed to segments without first validating the hypotheses. So, though you may think that soccer moms are the perfect segment for your creative and your offering, you can’t know for sure without testing first. Concrete data is always going to be more valuable than instinctive decisions!
Environmental data translates to specific keywords, the season, the weather, the type of connection, the date, the application category, etc. A good example is this ad from the Canadian donut and coffee giant. Pay attention to one thing – the creative is derived from a fixed template, which displays variations depending on the context. It would be counterproductive to create brand new creative for each segment!
This data takes the form of shopping carts, formats seen, past conversions, devices used, traffic sources, etc. It’s used to optimize the experience, and therefore engagement, and always to lead people to the next stage of the conversion funnel.
Now let’s get back to Avinash and his model. If you’ve followed up to now, you’ll have understood that it’s literally unthinkable to apply all the elements we’ve just explained on a large scale. Too many details, too many elements to control, too much data in your Excel spreadsheet. How can you manage five AMFM audience meta-segments, combined with the top/middle/low strategies from Avinash’s model, multiplied by the media environment, multiplied by your 15 offensives, all without losing control of your message? You’ll break your head trying to do too much.
The recipe for good optimization is simple: develop, iterate and control. One hundred thousand dollars for a 30-second television spot, transformed into a 6-second YouTube bumper used over a 2-month campaign… It’s expensive. A CMP, or creative management platform, aims first and foremost to save you money on production.
Some truths hurt.
More than 90% of campaigns use generic creative that doesn’t correspond to the desired audiences.
That’s where CMPs will respond to a niche need in the market. This technology develops a multitude of iterations from a simple base creative. Because of course that’s what costs the most in production. In the future, creative will no longer be a fixed file, but rather a folder of components that can be quickly assembled and reassembled.
All this to say: CMPs allow you to reach the right person at the right place in the consumer lifecycle.
Google is now offering a first version of a CMP, that allows you to work with its suite of tools to create, put online, host and purchase. It’s therefore important to really understand the inventory you’re buying as well as the existing purchase model.
The magic works through a well-defined audience strategy. If the audience says SPORT then the creative that’s displayed will be SPORT. If it’s TRAVEL then the creative would be TRAVEL. That’s why it’s important to have large and well-defined segments.
Several fundamental shifts in digital media have simultaneously simplified and complexified its processes. Let’s revisit the steps of this evolution.
In the beginning, in the Mad Men era, media was very simple. The advertiser called up their agency, who, old fashioned in-hand, communicated the need to the publisher, bought ad space, and bingo, the ad was seen by consumers.
Programmatic changed the rules of the game around 2006. Layers of intelligence and automation were inserted between the agency and the publisher, supply (SSP) and demand (DSP) platforms connected through an Ad Exchange. The result? We’re seeing better management of ad inventory, allowing us to sell more.
Then people started to realize that campaign management had become extremely complex, with its countless audience buckets. And that’s where the DMP (Data Management Platform) made its appearance, allowing both client and agency more automation. Except, all great advancements come with their share of challenges: advertisers now had hundreds of audience segments, and they were starting to be too much for agency campaign managers to handle.
Now let’s get back to 2019. Back to the CMP, possible saviour of modern times. This innovation allows us to manage this beast of complexity without relying on an Excel spreadsheet. Because too many spreadsheets, too many columns, too many numbers, ends in burn out. And regardless of your annual return on investment, burnout is always too high a cost to pay for the person who’s living it. As I mentioned earlier, you’ll break your head if you try to do too much.