Introducing “Digi-Log” measurement systems

Solution to measuring digital marketing effectiveness in Cookie-less world

Change in Google’s policy over not supporting 3rd party cookies and shifting away from precision-targeting of ads based on your internet behaviour has pushed marketers to find alternative ways of measurement and targeting.

These changes in the media and marketing world would mean a new approach for data acquisition, access & measurement. This approach has to be more privacy-centric and comply with new regulations & guidelines by using tactics including differential privacy that withholds information about individuals instead uses information that describes the patterns of groups within the dataset for public sharing.

Often chaos leads to transformation and that’s exactly what we are witnessing here. Our Data scientists & mathematicians have started testing their new measurement solutions and it proves to be more inclusive, strategic and futuristic compared to earlier solutions.

Ways to address measurability in Cookie-less world.

Until now, through digital measurement systems like- MTA, personalisation at scale without first party data was possible. The way multi-touch attribution answers this question is by taking into account the cost of each touchpoint and the weight (i.e.importance) you give at each stage of the customer journey. But it won’t be possible going forward and hence shift in measurement methodology is required.

Shift in measurement methodologies (digital-MTA to analog-Econometrics) due to change in available data grain (from individual cookie based data to aggregate cohort datasets) will impact data quality and introduce bias in measurement. To tackle this bias in measurement one deploys techniques like:

1. Panel-calibrated models
2. First party data based correction models
3. Co-create Universal ID spaces

Without cookies one does not have access to each individual’s path to conversion and thus one needs to go back to analog methodologies of econometric modelling for measurement purpose.

Analog methodologies are often looked upon as historical measurement techniques, as one would need at least 2–3 years of data to build relevant baseline and benchmarking models and estimates. Traditional econometric modelling is yet to prove its value and agility in these fast-changing ecosystems.

Yes, the question remains- will this work, with fast changing ecosystem and user behaviour?

I would say yes! If the econometrics is fused with almost real time, granular data, though at aggregate level, it will allow us the build multi ensemble structures, which will help us build connected systems and close the network by driving causality for the contextual systems.

We have tested and explored, the core principles of advanced analog systems (DIGI-LOG) and neo-classical econometric modelling techniques that enables one to measure the effectiveness of different advertising investments in driving the business KPIs, cross channel synergies and efficacy at aggregate level.

We are calling this DIGI-LOG systems: our Advanced analog measurement system frameworks.

This system includes–

1. Incrementality/ efficiency measurement framework, and
2. Conversion modelling & extrapolation framework. Here we deploy wide variety of advanced classical set of algorithms like:

2.1 Contextual causal probabilistic integrated models: Dynamic multinomial MMM models to estimate the long‐term and short‐term effects of marketing mix variables on brand KPI.

• Causal Structural Models: Impact and Synergy Analysis
• Cohort segmentation models: Audience attribute clusters
• Measurement Models- Regression & Time Series variants: Forecasting, Simulation and Optimisation

2.2 Advanced Neo-classical fusion models:

• Stochastic and Deterministic combined approaches (Bayesian and Space models)- Simulated Optimisation models borrowed from Financial, Biochemical Stochastic & Deterministic frameworks
• Ensemble Models- Demand based utility model and its industry variants integrated with real time aggressive and passive MMM data models

These advanced multi-domain designs, simulations and state of art econometric & measurement constructs have been tested and proven to use the pseudo-analog real time ensemble systems with even 3–6 months of historical data to predict future at 90%+ accuracy.

it’s indeed a “DIGI-LOG” era (Digital + Analog) where existing algorithms and theorems need to be tweaked to re-align to this ever-changing marketing ecosystem. We think Digi-log is the answer to measure the effectiveness of your digital marketing investments in the post cookie world.