Digital marketing trends for 2021.

In a highly connected environment get to know what the main trends are for 2021.

Data engineering, data science, cross-device analytics and mobile first, are ideas that have been spreading in the digital marketing world, in this article we will discuss the major trends in digital marketing for 2021.

The increase of complexity is a common issue in the digital world, not only the devices are evolving with new functionalities every day, but the very technology tools that are part of the marketing field are evolving too. In the digital marketer routine, some buzzwords are quite common, such as: Google Ads, Google Analytics, Firebase, Attribution Model, Conversions. Just to name a few. In the last 5 years data science has become popular, it basically stands that all data collected through marketing campaigns, analytics tools and internal data can be used together to get rich insights about users’ behavior and they can even predict what users are about to do. Along with data science let us discuss other hot topics that will generate buzz this year.

The digital marketing goal is basically to show ads to users in the form of pictures, videos, texts and e-mails that can influence them to visit a website, a Facebook page, to download an app or to make an online purchase. Not too complicated to understand, the same marketing principles and objectives are applied to smartphones, televisions, smartwatches and many others. But digital marketing has a very particular characteristic — it can be measured, which means that you can know, count and report the number of users who have seen an ad while navigating on Instagram, you can know how long a user spends on a bike purchase or even which country and hour of the day is more prone to have subscriptions. The field has been evolving and now let us see about this year major trends.

Cross-device tracking.

Since users can connect to any application from several devices or even buy a new smartphone to do it, it is crucial to keep tracking them despite the device change. We can think of a journey that starts on the notebook, when a user is searching for a new car model, it could take a while until he or she can finally decide which car model to buy and finally purchase it. The process of buying a new car is usually longer than buying a new bed, it makes sense, right? The thing is that long journeys could show more about your customers’ behavior, decisions and preferences, and they could occur in multiple devices. The key to be able to do cross-device tracking is to collect a user’s identifier, it could be any information since it is not a PII (Personal Information Identifier), such as a hashed e-mail address. Once you have all the information stored in your databases you can gather them with the key and understand the complete timeline of your customers.

Mobile Attribution and deeplinks.

It is no news that mobiles have overcome the desktop access. So, do you operate your marketing campaigns keeping that in mind? Traditional links redirect users to websites and beautiful landing pages, which is not wrong but deeplinking is definitely a better choice. Deeplinks can open the pre-installed app directly in the screen you want, so the journey becomes as short as possible since no redirecting to websites is needed. It could be tricky and complex to implement but it is worth the effort, App Links and Universal Links are the best choices for Android and iOS, respectively. You should run the deeplinks along with a MMP (Mobile Measurement Partner) such as AppsFlyer or Adjust, to have a full attribution visibility from the installation to the KPI’s.

Custom attribution models.

For those who have some experience in the marketing field know that each advertiser has the tendency to inflate its own numbers. The diversity of attributions models (last-click, last non-organic click, linear) combined with the limited information each advertiser’s pixel can obtain, makes the reports of your investments not accurate as it could be if you had a custom attribution model. Suppose that a user clicks in Ad ‘A’, ‘B’ and ‘C’ and a conversion is followed, custom attribution modelling is the ability to decide how to weight each of the touch points as your convenience fine-tuned with your own business requirements. Although it requires a lot of data engineering effort and granularity of user level information it delivers the best reporting as possible.

Custom audiences creation.

The investments in digital marketing could reach to millions of dollars in big companies, but not only the big ones suffer with bad marketing strategies. Startups that have limited resources must target the right audience so to not waste any money. The ability to create custom audiences that will be reached by your campaigns, not only saves you money but also guarantees the affinity with your brand, preventing flooding ads to those who do not want your products and services. Ok, but how to create custom audiences? Each advertiser tool: Google Ads, Facebook Ads, LinkedIn and others have their own process to import list of users, it could be a .csv file with e-mail addresses or telephone numbers, for example. If the audience you are creating is from mobile consider using the Advertising ID to reach the users. Advertising ID is a unique device, and it is as accurate as possible to identify your user without custom IDs. Keep in mind that you should have a CRM database or analytics tool already running to be capable to list the users, if you can use statistical analysis to enrich the audience with look-a-like behavior you are doing just fine!

Data Engineering.

In the topics of “custom attribution model” and “custom audiences” discussed above all of them are part of data engineering. Data engineering is the creation, organization and processing of multiple data sources to get rich analysis. In digital marketing scenario the sources are often analytics tools, advertiser data and CRM. The area is formed by several technologies: Big data services such as Amazon S3 or Google BigQuery, along with data processing frameworks as Apache Beam or Spark often running on powerful servers in cloud. That is where the most operational activities happen to get business level reports that really matter.

Data Science.

Once there is mature data collection and structuring it is time to go to the next marketing step. Data science makes it possible to cross and correlate data in new ways humans would hardly be able to. Did you know that beers seem to be correlated with diaper sales? It is an association that machine learning can do by observing and training statistical models in sales data. In order to get good and reliable insights from machine learning, it is needed to have people that understand both the data science tools and processes as well the business characteristics itself. Data visualization tools such as Tableau and Data Studio make it possible to easily understand all the magic of predictions in a graphical way. Python is a programming language where very interesting open-source data science solutions were born, and it seems to be a technology that will last for at least a few years, to explore and study hidden relationships from standard data. In digital marketing the power of data science is to cluster users in many dimensions as interests, location and purchasing power. The clusters created could be used to target high effective marketing campaigns — showing the ads to the right consumer at the right time.

Privacy politics (iOS 14) and cookies policy updates.

The tech industry is passing through a moment that the privacy and data collection are constantly being debated and suffering changes. The cookies restriction was the first implemented decision in order to avoid user tracking, preventing users’ navigation information to be sent to analytics tools. But it seems to be just the start, with the new privacy updates of iOS 14, the apple devices will no longer share its own device identifier which means the companies and advertisers won’t be able to see the actions of an individual user from the beginning to the end and cross that information with other data sources outside the app vendor. For the users it is a big step from preventing undesired notifications and marketing ads but for marketing industry it is a huge problem. It will change almost everything from the data visibility to the number of campaigns you can run optimized by conversions. The IDFA will no longer be collected — unless the user accepts it - another identifier, the IDFV (Advertiser for Vendors) is still available besides its limited scope. The advertisers have been presenting webinars and discussions to share the new recommendations and practices that Apple has enforced them all.

It is a pleasure to share these insights with you and I hope that it can be useful!

Rodolfo Luis Marcos - Data Engineer and tech Lead.

[email protected]