Data and the danger of 'creative-washing'

By Veronica Patton-Cemm, Consulting Director, Insights

This year’s Cannes has been a return to form, replete with amazing campaigns and robust discussions about the future of creativity and communications. Award winning campaigns have showcased impactful outcomes for war refugees (Room for Everyone by Mastercard), corruption & transparency (Transparency Card by Congresso em Foco), and pub preservation (Pub Museums by Heineken) alongside more light-hearted campaigns from companies such as Dramamine.

Across the winning cohort, there’s been a noticeable focus on data, evidence and demonstrating impact, with brands and businesses keen to showcase that their campaign has truly connected with audiences by highlighting a societal issue and harnessing data to catalyse change (see: Transition Body Lotion by Vaseline). The use of data as an input – the basis of the campaign – is a necessary one. In many respects it acts as a bridge for businesses. How does a business identify issues that are important to their audiences and stakeholders? What can data tell us about the scale of the issue and how it effects society? And how can data be used to connect issues with company purpose? Data is the answer but its more nuanced than that. How data is used to underpin and demonstrate the impact of campaigns is a delicate and deliberate balancing act. Biased thinking, flawed methodology and misinterpretation of data, such as confusing causation and correlation, can lead to skewed conclusions and incorrect outputs.

In our review of the winning works, we delved into how businesses utilised data and insights to spark conversation. One of our favourite campaigns, the Pink Chip, makes extensive use of data to re-frame the conversation around female leadership. However, the data it uses to do this and the methodology underneath it raises some interesting questions.

The campaign to overcome bias towards female CEOs made use of four years of data to definitively state that female CEOs are better for business. The re-framing of the dearth of female CEOs as an economic oversight instead of a societal issue helped bring new attention to the on-going challenge. And it is a significant issue which needs to be highlighted and addressed – female under-representation in leadership positions and the inherent misogyny women deal in the workplace is a global challenge. Recent statistics reported continued improvements in the number of female corporate leaders amongst Fortune 500 and FTSE-listed businesses. Yet we know there’s still further to go in achieving true gender parity in business.

So, the findings shared in the campaign resonate strongly by combatting stereotypes and harmful ideas about the effectiveness of women in leadership positions. Yet upon further review, we noted a few issues that could potentially undermine the strength of the insights.

There is a continued challenge in presenting quantitative data using a small sample, particularly when the result and messaging feels pre-determined. This introduces bias into the work, specifically confirmation basis. There are many ways to combat inherent bias including weighting and introducing controls for other considerations and factors. In this case, there’s a need to balance the findings across external and sector-specific factors. An example a senior colleague posited was the Australian oil and gas company Woodside US O&G company Occidental both have female CEOs, and both have significantly outperformed the index average. Yet so have Exxon, BP and Royal Dutch Shell as well as other oil and gas companies that do not have women CEOs and are not included in this index. This demonstrates how big the sectoral aspect is when reviewing the findings.

Creating like-for-like comparators means stripping away the ‘noise’ and ensuring alignment across data points. In this case, ‘noise’ includes foreign exchange rates and inflation. It may also mean accounting or creating ways to mitigate the impact of corporate actions that could impact the comparator data points which the study did and clearly outlines in its methodology.

Taking it further would mean adjusting sector by sector to reflect ‘a rising tide that raises all boats’ factors as well as the inverse of that, so that in the end the only difference between comparator companies is the gender of their CEO.

It is encouraging to see the harnessing of data and evidence to great effect, and these considerations will likely not affect the impact of the overall message and its importance. Data, evidence and insights will continue to play a significant role in the development of winning campaigns, particularly in the use of AI to gather information. However, when data underpins the work, it must be robust and weighted to both identify and remove bias and create a strong foundation for benchmarking and future work that delivers impact.

Stuart Lambert