Digital marketing growth faces three stubborn headwinds: measurement blind spots, ad-cost inflation & algorithm volatility, and the personalization/privacy paradox. Each has pragmatic fixes.
Blind Spots of Measurement: Cookie-based cross-site tracking continues to decline due to increased privacy laws and regulations leading to the breakage of most legacy attribution models used by advertisers and publishers. This has had a substantial impact on the ROAS calculations and programmatic targeting of advertisers and publishers alike. To temporarily address blind spot issues, it is suggested that companies begin incorporating both first-party measurement efforts, as well as implementing privacy-preserving methodologies (MMM, testing for incrementality) while negotiating more concerning partner data collaboration. Over the medium term, companies are encouraged to utilize identity solutions (universal IDs), where permitted by law and ethically.
High ad costs & volatility in platform auction pricing are causing increased cost of advertising and unpredictable reach from continuing to pay for ad spots. Businesses need to stop relying on just the paid side of advertising. A better approach is to divert more of your advertising dollars into your own channels (email, text/SMS and Organic Social channels), use content and assets on demand using a modual content creation model and the use of Creative Testing in the Digital Ad Buying Process to create more relevance to your audience and, in return, reduce the cost per thousand impressions (CPM).
There is a higher conversion rate with Personalization, however, there are many ethical and regulatory compliance risks associated with it. To mitigate those risks, build-out Explicit Consent First Flows using Exchange Benefit programs. The program will provide value to consumers for sharing their personal information (better prices on future purchases, faster checkouts), conduct regular Privacy Audits.
Lastly, all companies should use AI in a responsible manner. They must be vigilant about avoiding Bias in AI and if at all possible utilise Collective / Synthetic Data for training their AI Models.