Self-service AI refers to the intelligence that business users (analysts and executives) can acquire on their own from the data without the extensive involvement of data scientists and engineers. It means enabling them to acquire the actionable intelligence to serve their business needs by leveraging the low-code paradigm. This results in reduced dependency on other skills such as core IT and programming and makes faster iterations possible at the hands of business users.

As the data inside and outside of the organizations grows in size, frequency and variety, the classical challenges such as hard-shareability across BUs, lack of single-ownership and quality issues (missing data, stale data, etc.) increase. For IT teams owning the data sources, this becomes an additional task to ensure provisioning of the data in requisite format, quality, frequency and volume for ever-growing analytics needs of various BU teams, each having its own request as a priority request. Think of the several dashboards floating in the organizations created at the behest of various BU teams, and even if with great effort they are kept updated, it is still tough to draw the exact insights that will help take direct actions based on critical insights and measure their impact on the ground. Different teams have different interaction patterns, workflows and unique output requirements – making the job of IT to provide canned solutions in a dynamic business environment very hard.

Self-service intelligence is therefore imperative for organizations to enable their business users to take their critical decisions faster every day leveraging the true power of data.

Enablers of self-service AI platform – Incedo LighthouseTM

Incedo LighthouseTM is a next-generation, AI-powered Decision Automation platform targeted to support business executives and decision-makers with actionable insights generation and their consumption in daily workflows. Key features of the platform include:

  • Specific workflow for each user role: Incedo LighthouseTM is able to cater to different sets of users, such as business executives, business analysts, data scientists and data engineers. The platform supports unique workflows for each of the roles thereby addressing specific needs:
    • Business Analysts: Define the KPIs as business logic formulations from the raw data, also define the inherent relationships present within various KPIs as a tree structure
    • Data Scientists: Develop, train, test, implement, monitor and retrain the ML models specific to the use cases on the platform in an end-to-end model management
    • Data Engineers: Identify the data quality issues and define-apply remediation across various dimensions of quality, feature extraction and serving using online analytical processing as a connected process on the platform
    • Business Executives: Consume the actionable insights (anomalies, root causes) auto-generated by the platform, define action recommendations, test the actions via controlled experiments and push confirmed actions into implementation
  • Autonomous data and model pipelines: One of the common pain points of the business users is the slow speed of data to insight delivery and further on to action recommendation, which may take even weeks at times for simple questions asked by a CXO. To address this, the process of insights generation from raw big data and then onto the action recommendation via controlled experimentation has been made autonomous in Incedo LighthouseTM using combined data and model pipelines that are configurable in the hands of the business users.
  • Integrable with external systems: Incedo LighthouseTM can be easily integrated with multiple Systems of Record (e.g. various DBs and cloud sources) and Systems of Execution (e.g. SFDC), based on client data source mapping.
  • Functional UX: The design of Incedo LighthouseTM is intuitive and easy to use. The workflows are structured and designed in a way that makes it commonsensical for users to click and navigate to the right features to supply inputs (e.g. drafting a KPI tree, publishing the trees, training the models, etc.) and consume the outputs (e.g. anomalies, customer cohorts, experimentation results, etc.). Visualization platforms such as Tableau and PowerBI are natively integrated with Incedo LighthouseTM thereby making it a one-stop shop for insights and actions.

Incedo LighthouseTM as self-serve AI at a Pharmaceutical Clinical Research Organization (CRO)

In a recent deployment of Incedo LighthouseTM, the key user base is the Commercial and Business Development team of a Pharma CRO. The client, being a CRO, had drug manufacturers as its customers. The client’s pain point revolved around the low conversion rates leading to the loss of revenue and added inefficiencies in the targeting process. A key reason behind this was the wrong prioritization of leads that have lower conversion propensity and/or have lower total lifetime value. This was mainly due to judgment-driven, ad-hoc and simplistic, static, rule-based identification of leads for the Business Development Associates (BDA) to work on.

Specific challenges that came in the way of application of data science for lead generation and targeting were:

  • The raw data related to the prospects – using which the features are to be developed for the predictive lead generation modeling – were lying in different silos inside the client’s tech infrastructure. This led to inertia to develop high-accuracy, predictive lead generation models in the absence of a common platform to bring the data and models together.
  • Even in a few exceptional cases, where the data was stitched together by hand and predictive models built, the team found it difficult to keep the models updated in the absence of integrated data and model pipelines working in tandem.

To overcome these challenges, the Incedo LighthouseTM platform was deployed that allowed them to:

  • Combine all the data sources’ information into a Customer-360-degree view, enabling the BDAs to look at a bigger picture effortlessly. This was achieved by pointing the readily available connectors within the Incedo LighthouseTM platform to the right data sources, and establishing data ELT pipelines that are scheduled to run in tandem with the data refresh frequency (typically weekly). This allowed the client’s business analysts to efficiently stitch together various data elements, that were earlier lying in silos, in a self-serve model and include custom considerations that are region and product specific during the data engineering stage.
  • Develop and deploy AI/ML predictive models for conversion propensity using Data Science Workbench which is part of the Incedo LighthouseTM platform, after developing the data engineering pipelines that create ‘single-version-of-the-truth data’ every single time raw data is refreshed. This is done by leveraging the pre-built model accelerators for predictive modeling, helping the BDAs sort those prospects in the descending order of their conversion propensity, thereby maximizing the return on the time invested in developing them. The Data Science Workbench also helped with the operationalization of various ML models built in the process, while connecting model outputs to various KPI Trees and powering other custom visualizations.
  • Deliver key insights in a targeted and attention-driving manner to enable BDAs to make most of the information in a short span of time. This is achieved through well-designed dashboards to rank-order the leads based on the model reported conversion propensity, time-based priority and various other custom filters (e.g. geographies, areas of expertise). The intuitive drill-downs were encoded using the region-specific KPI Trees to enable them to know the exact account portfolios of their business that were lagging behind. These KPI Trees were designed by the client’s business analysts within the platform’s self-serve KPI Tree Builder, saving multiple iterations with the IT teams. The KPI Trees allowed the BDAs to double click on their individual targets, understand the deviations from actuality, and review the comments from earlier BDAs who may have been involved, to decide the next best actions for each lead.

The deployment of Incedo LighthouseTM not only brought about real improvements in target conversions, but also helped transform the workflow for the BDAs by leveraging Data and AI.

Ashish Gupta
Ashish Gupta
Head of Data & AI
Recruitment Fraud Alert