Listen to the podcast in French on

.

2021 was a year of transition marked by a strong increase in data maturity. This podcast explores trends to be expected in the months to come in regards to accelerating 1P data collection, expanding brands’ CDPs and increasingly probabilistic approaches to performance measurement.

2021 was a year of transition marked by a strong increase in data maturity for:

  • Companies: Data is becoming absolutely strategic: 90% of companies have recruited in data according to a survey by the “mission numérique gouvernementale des grands groupe” (governmental digital mission of large companies)

  • Citizens and consumers: There is a growing awareness of the importance of data protection (we saw it this year highly publicised reports on the dangers of fake news and the effects of echo chambers)

  • Regulations & governments: GDPR continues to create ripples around the world, in the US with the CCPA in Virginia (VCPA) and Colorado (CPA), and even in China with the PIPL, enforced since November 1st

The trends we are observing are part of this series of changes, with advances in technology combined with greater respect for data protection

The need to control one’s 1st party data, as it is becoming vital to be able to rely on clean, consented and controlled data. This requires two enablers: data collection programs and a proper tools ecosystem

  • Accelerate on 1P data collection:

    1. 3rd party data is on the way out: the three major browsers have or will eliminate the use of 3rd party cookies, so we expect more efforts to collect large amounts of 1P data via more innovative programs.
      • For example, McDonald’s in France has made its annual Monopoly campaign a 100% digital program, requiring the creation of an account on the McDonald’s+ app where you must scan your stickers to find out your prize.

  • The CDP (consumer data platform):

    1. Brands need an extensive CDP framework to collect, store, and manage audiences using consumer data, both PII and non-PII. Currently, no tech solution on the market can cover 100% of the ambition, so we have to make do with assembling multiple tools. We rely heavily on the Google stack and we have built an offer that relies on Google Marketing Platform as the main stack that we complete with technologies such as Tealium, Treasure Data and many others
    2. On collection: It is becoming essential to use “data clean rooms” to ingest and process media data in order to respect privacy and consent: Google ADH, Amazon Marketing Cloud, FB Advanced Analytics…
    3. Last point: We think that some critical functionalities must be built internally. For instance we have launched a method called Audience Engine, which allows you to manage your own audiences on top of the CDP to avoid relying on the exact same models as your competitors (we do this for major brands such as Reckitt or Samsung)

A concrete application of these changes is the evolution in performance measurement:

Historically, we have relied heavily on deterministic measurement based on identifiers (mainly 3rd party cookies) to track the consumer journey from start to finish online. The end of 3rd party IDs will quickly make traditional attribution approaches less relevant. They will continue to work on a reduced scope, within walled gardens (you will have your FB performance measurement, attribution for the entire Google environment – cross-device thanks to GA4) but will be less effective for a cross-walled garden vision.

So we expect an increase in probabilistic approaches, which consists in linking historical performance mathematically to connect sales with marketing actions. These have the immense advantage of being exhaustive (taking into account all sales and all marketing/sales actions).

  • However the approach is still not perfect:

    1. Issues with granularity – hard to have relevant insights at a campaign level for instance
    2. Issues with complexity: models rely on such large datasets, it takes anywhere between 3-6 months to update them
    3. Explainability: The main challenge here is these models are largely based on historical correlations between sales and marketing/media expenses which do not imply causality – there is a strong correlation between the number of annual pool deaths in the US and the number of Nicolas Cage appearances in a movie – but rather hard to find a causal link there…

  • To solve these issues, Artefact has opened a research centre in partnership with Google and Boston University to work on new probabilistic models. We have tested some very promising approaches, in particular with Reckitt Benckiser, using Bayesian networks, which are built on probability analysis rather than linear regression, able to handle very granular data, are quick to train, and are very understandable by business teams without advanced knowledge in data science.

Convergence between sales & data

After years of talking about it, this is the first time that we effectively see a convergence of these teams. This is already very true in Asia where we have clients for whom the E-com channel represents up to 60% of sales. The particularity there is that platforms like T-Mall and JD.com include both the traditional retail negotiation (sales conditions, promotional investments, assortment evolution) AND the media investment. This means that brands must integrate not only SALES teams, but also MEDIA/MARKETING teams, and more and more DATA & TECH skills in their discussions at the same time, hence the need for convergence

The first use cases behind this convergence are primarily in retail media

  1. (using a brand’s media budget on the retail space to promote brands with very strong use of data to target customers or prospects stored in the retailer’s fid base) We are also expecting a strong acceleration of online trade marketing: Indeed, the trade and media budgets are the same size, but where the digitalization is 50% on the media, it is barely 10% on trade marketing! We expect a strong adjustment on this front.
  2. This will be possible thanks to an explosion of “Retail Tech” technologies: Links with Carrefour (where Artefact is one of the official partners and operates campaigns for Unilever brands) which uses Liveramp and Google technologies, Relevancy for Casino, Walmart Connect…
  3. Beyond retail media, we see other very interesting brand-retailer collaboration use cases appearing, for example around Category Management such as personalised promotion, management (assortment), price control or supply chain integration with a VMI logic where tomorrow the brand can operate part of the retailer supply chain.

Conclusion

I would like to conclude on the data culture in companies

  1. Today, we see a significant improvement in the maturity of companies on these subjects. The “core” teams are now in place and it is time to train the rest of the organisation. At Artefact this is what we call data democratisation
  2. According to APEC, the demand for data experts, experts in data manipulation, understanding and measurement, is exploding: +x2 for DS/DE , for DAs x4 between 2020 and 2021
  3. It is to meet this need that we have created our Artefact School of Data, in particular to help people who are in a situation of professional reconversion, to find opportunities in the data sectors

Listen to the podcast in French on

.

Artefact Newsletter

Interested in Data Consulting | Data & Digital Marketing | Digital Commerce ?
Read our monthly newsletter to get actionable advice, insights, business cases, from all our data experts around the world!

Newsletter Sign UpNewsletter Sign Up