Redefining Digital measurement-From Web analytics to Digital Business Intelligence
With the proliferation of digital channels and the increasingly complex multi-touch consumer journeys, the web has come to play an increasingly pivotal role as a centralized data store containing a goldmine of information about consumer behavior, context, wants and needs. Despite the significant advancements in tracking technology and introduction of visitor level tracking techniques, contemporary web analytics tools remain-at best-rudimentary mechanisms for recording click stream behavior and basic activity based metrics. Largely devoid of ‘analytics’ capabilities as we come to understand them today (advanced pattern recognition, cause and effect analysis, clustering etc.) or even basic business intelligence features (multi-dimensional drill down reporting on cross channel data, ad-hoc analysis of large historical datasets etc.), the tools provide a good enough technology platform for early stage Marketers looking for basic information about website activity
Our position is that a serious digital insights strategy should explore implementing a well architected digital business intelligence solution in order to truly unlock the potential of web channel data. Formulating and tracking business relevant metrics that go beyond basic ‘counting’ of website events to ones that make effective use of cross channel data for marketing measurement leads to significantly better utilization of analytics investments than conducting random, ad-hoc analysis within web analytics tools alone.
Migrating from Web Analytics to Digital Business Intelligence
- Avoid sending offline data into web analytics tools
- Clearly separate the data collection, storage, processing and reporting functions
- Define metrics upfront and develop a top-down data strategy for tracking them
- Design and implement a strategic measurement technology roadmap that evolves incrementally with measurement needs
However, using digital business intelligence as a substitute to ‘web analytics’ is not being recommended either. We simply posture that greater focus should be put on upfront metrics planning and on creating a robust data and measurement technology strategy that supports tracking of those metrics in order to truly understand the wider issues around operational performance, customer profitability, customer service and overall business growth
Digital Marketing Measurement
3 categories of metrics
Web Analytics tools hold a large amount of behavioral data but using them unscrupulously to track advanced, cross channel marketing metrics or even for tracking operational performance results in largely wasted implementation effort with little outcome beyond low value static reports
Excessive slicing and dicing of native data on the other hand often results in analysis paralysis situations with questionable business value
Before discussing the blueprint of what a digital business intelligence solution could look like, here is a quick outline of the three high level categories of metrics commonly tracked as part of marketing intelligence initiatives
- Count or Activity based metrics-In short, these answer questions around what are we doing. Number of conversions, conversion rates, rate deltas, bounce rates, content page views, trends etc. all address questions around how many events are taking place and/or how is the trend changing over time
- Operational performance metrics-These metrics bring the cost, and productivity aspects of Digital Marketing into play. Generating 100 conversions in a day may not sound like an achievement anymore if the cost per conversion exceeds 300% of budgeted allowance or if the long term value generated from these conversions far exceeds their acquisition costs. While activity based metrics answer questions around what are we doing, operational performance metrics provide insights into are we doing it right. Channel ROI, Campaign ROI, Lifetime value, Product/Category margins are all examples of commonly tracked operational metrics. From a technical perspective, operational metrics involve combining data from multiple sources over differing periods of time and are more challenging to track than count based metrics
- Business outcome metrics-Knowing the number of conversions and even the ROI of investments brining them in is still not enough when taking a strategic view of marketing. Simply generating x conversions within stipulated ROI mandates tells us nothing about strategic business outcome metrics such as long term customer profitability, NPS scores, market share, share of voice etc. In other words, doing something right does not necessarily amount to doing the right thing! From a technology perspective, business outcome metrics typically use multiple data points (in many cases their data resides outside organizational boundaries) and are significantly more challenging to construct
In view of the metrics continuum outlined above, we can now explore the architectural options for implementing a digital business intelligence solution
OPTION 1- TURNING THE WEB ANALYTICS TOOL INTO A BUSINESS INTELLIGENCE APPLICATION
Effectively, this option requires building interfaces to send all data into the Web Analytics tool. Most tools including Google Analytics, Adobe Analytics etc. tout features that allow uploading of a variety of offline data into their tools (cost data, SAINT classifications etc.) but the architectural merit of these options is severely suspect for following reasons
- The data model used by Web Analytics tools is primarily designed for storing aggregate level, click stream data. Even with tools that support visitor level analytics, the data model is fixed in stone making it pretty much useless for tracking advanced operational and business outcome related metrics that require flexible data models designed for specific business scenarios. In any case, a standard practice in most commercial deployments is to pull data out of the Web Analytics tool and do reporting in Excel or some other data analysis tool. Why then send data into Web analytics in the first place?
- Business intelligence as a function implements multiple capabilities including ad-hoc analysis of large datasets, multi-dimensional data analysis, advanced visualization patterns, support for customized data models, access to individual level data, collaboration, granular access control, single sign-on integration, support for unstructured data analysis and many more. Most of these features are well beyond the capabilities of most current Web Analytics tools
- Business intelligence extends well beyond Marketing in most companies that look to track advanced operational and strategic business outcome metrics. Using an existing BI setup for analyzing digital channel data allows Marketers to reuse a number of pre-built capabilities (e.g. security, role based access, raw data access, ETL capabilities etc.) thereby significantly reducing time to insights
- The web represents just one of the channels involved in a modern day consumer journey that spans multiple digital and offline channels. Fueling web channel monopoly by sending in interaction data from all the other channels makes little sense politically and zero sense technically
OPTION 2- Using a dedicated business intelligence solution
The challenges outlined above can be mitigated to a large extent by using a dedicated Business Intelligence solution-ideally one that is architected for Digital Marketing use. A basic, layered architecture for such a solution would consist of 4 layers
- Hosting and infrastructure-This consists of all components required to host the BI solution in line with non-functional requirements around security, uptime, role based access, portal integration, single sign-on etc. In addition, this layer would include the deployment architecture (development, staging, production etc.) configuration to ensure seamless insights delivery
- Data storage-This layer consists of all the logical data models designed for specific analysis requirements along with physical data stores that might be best suited for particular use cases. For example, near real-time analysis of in-flight campaign data may require in-memory computing engines but if the requirement is limited to end of period (week/month etc. )data analysis, then disk based databases might suffice. In scenarios that involve large number of dimensions, adopting a data warehousing technology might be a better option than using RDBMS stores
- Data Integration-The data integration layer consists of physical ETL tools and architectures required to implement the various interfaces. Typically the ETL tool would consist of out of the box connectors for various peripheral systems in scope while the deployment architecture would consist of components for implementing non-functional requirements such as software reuse, error handling, automated ETL job deployments, job monitoring etc.
- Reporting Frontend-This layer consists of all the end user focused components such as the reporting tool (e.g. Tableau, Qlikview, Microstrategy etc.) along with provisions for implementing advanced features such as KPI alerts, data profiling, forecasting etc.
4 layers of a Digital business intelligence solution
Decoupling the ‘Business Intelligence Solution’ into layers outlined above allows Marketers to build their own technology stack from scratch or plug into wider corporate level technology components already deployed. From a business point of view, using a dedicated Business Intelligence solution for digital data analysis allows Marketers to significantly improve the quality of their analysis output and thereby provide bottom line improvements in Marketing performance
Interested in finding out more about Digital Business Intelligence Solutions? Please check out STRATIFY-our proprietary architecture development framework designed specifically for Digital measurement. With over 16 architecture building blocks organized into 4 layers, STRATIFY allows Marketing Technologists to quickly assemble digital measurement technology blueprints in line with business priorities and resource constraints