Where data meets creative: 6 marketing transformation lessons from the frontline

Anastasia Leng October 2018 Data & Measurement

In almost any area of any industry, being “data-driven” is seen as a plus. But data-driven creative is still unusual, and many companies are yet to embrace its potential. That’s all about to change, says Anastasia Leng, founder and CEO of Picasso Labs, former Googler and pioneer in the field. With insights, advice and predictions for the future, Anastasia shares her thoughts from the marketing transformation frontline.

When I co-founded an e-commerce company in 2012, it didn’t take long to recognize how important imagery and video was to almost every customer purchase decision. But as certain as we were that visuals were vital, what made some more successful than others? We went to our data and began puzzling out the impact of visual cues on customer behaviour. How should we quantify visual preferences? Google’s search campaigns approach different audience segments with different keywords. We looked to do the same, using images instead of words.

In 2015, that project became its own company, Picasso Labs, and in 2017, we took it to market, attracting clients from Unilever, Samsung, Deliveroo and Farfetch. We discovered that creative was often the only part of a brand’s marketing funnel where data was not being used in a systematic way. Marketers would know what time of day to approach customers for maximum impact, but knowing what to show them was largely guesswork.

Data-driven creative strategies deliver transformative performance improvements on CTR, views and conversions. They optimize media spend, too, thanks to clearer briefs and more targeted commissions. The end benefits are clear, but switching to data-driven creative brought its fair share of surprises, too. Here are six discoveries we’ve made in our journey so far.

#1 Every company has creative myths

We knew that a lot of decision making on creative was subjective, but we’ve been amazed again and again by how idiosyncratic it can be.

In one example, we began working with a major CPG client that prides itself on being data-driven. We started off by trying to assess how much the data we’d assembled from their visual creatives matched up with their expectations as a team. So we took one of their beauty products and asked: Does the hair color of a model affect the performance of ads for this product?

Their reply was emphatic: absolutely it made a difference. So we asked which hair color was preferable of blonde, brunette, redhead and black hair. About 90% of the people in the room, some 50 of whom were senior marketers on this product, raised their hands for redhead. We asked why, and they said it’s the most unusual hair colour, so it captures people's attention better than anything else.

In fact, our data indicated that redhead was the worst performing hair color of the four in all markets bar one, and it was actually the least effective for this brand. We probed deeper to understand where this misconception had come from, and we discovered that in the past a very senior creative director at the company had insisted on including a redhead in every campaign. A consensus had built up around that within the company, with a negative impact on years of creative work.


"Data helps creatives incorporate brand guidelines and philosophies so that they can see the space they have to experiment."

- Anastasia Leng, Founder & CEO, Picasso Labs


#2 Challenging gut instincts is hard, but worth it

One of the biggest problems for creatives and marketers stems directly from the absence of factual information on performance. Without metrics to refer to, debates about what will be effective rapidly boil down to questions of opinion. In place of a nuanced, informed discussion, creative approvals become a simple question of “I like it,” or “I don’t like it.” That’s frustrating for creatives, inefficient for marketers and unproductive for the process.

Creative should be audited by consumers, not marketers, using concrete, granular data from the way media is received in the real world. We’ve found that for many companies, that’s not an easy change to make. It can be hard to accept that gut instincts, which have guided major creative decisions for so long, can be wrong.

#3 Creatives need data to make their case

When a creative walks in to pitch an idea, they are usually the least-informed person in the room. All around them are marketers armed with reams of information – from media people who live in spreadsheets day and night to insights managers with data at their fingertips. Standing in front of them, the creative presents ideas that have taken weeks to prepare, and the marketers can simply refer to their data and say “I don’t like it.” Because creatives can’t respond in a discussion over data, their proposals can be punctured with a few crisp, sharp facts.

That’s not the case for anything else that a marketer does. If someone proposes a strategy to optimize keywords, for example, they will back up their ideas with data and be ready for a discussion about their interpretation of the facts. For creatives, that process has been much more fragile, and it doesn’t need to be.

Creatives using data can explain why their decisions are more than a matter of taste. They can point to evidence and make a robust case for their proposals, explaining why specific details will lead to improved performance and better customer responses. They can defend their ideas on a level playing field, with science to back up their intuition.

#4 Data helps align brand identity, and machine learning is next

It’s a given that aligning creative is key to establishing and maintaining brand identity, but for large companies, that can mean customizing creatives across hundreds of products in multiple markets. We’ve seen campaigns with over 200,000 variations of its creatives worldwide. How does a creative director keep branding consistent across so many iterations? Data helps creatives incorporate brand guidelines and philosophies so that they can see the space they have to experiment. Machine learning will be the next step, generating a clear framework for creative to help directors keep track of variations within the brand.

#5 It’s all about interpretation, so don’t be too literal!

We recently did some work for a multinational food and beverage company, applying data to the creatives marketing beer. We discovered that when audiences could see a hand holding the beer, instead of the beer on its own, the performance of their adverts increased in every market, across every single brand.

Were hands the secret to higher sales? Should every image now include at least one hand? The company considered the data and decided that hands alone were not the answer, but that a human touch really makes a difference. It challenged its creative directors to think about how that human touch could be included in their creatives.

#6 Data-driven doesn’t mean data dictates

Data-driven creative shouldn’t be about pushing a prescriptive approach, it should be about helping creatives build on what they’ve done before, defend their ideas, and produce work which really delivers. A data-driven approach can never be a replacement for fresh-thinking. Creative is about imagination and interpretation, taking nuggets of information and transforming them into brilliant ideas that people really respond to. Data-driven creative doesn’t mean creative is going away. It means its going to get better.

Why leadership is the new focus for digital transformation