Case Study | Improving data reliability and accurate insights through AI and predictive analytics

Hear how a leading research and consulting organization turned to Amplifi to implement predictive analytics and a private, bespoke LLM to achieve core business goals.

“We need AI”

“We need predictive analytics”

“We need a single view of customer”

These are all requests that we regularly come across, and it always takes a bit of investigation to uncover what organisations want, what they need and what value it is expected to bring.

In mid-2023, a leading research and consulting organisation, dependent on subscription revenue, turned to Amplifi for help. The organisation needed to implement “predictive analytics” but faced significant challenges with unreliable and poorly understood data.

The organisation had a vast amount of data related to customers, publications, conferences, and consulting services. However, the data's reliability and comprehensibility were major issues, hindering their ability to leverage advanced technologies like Generative AI, machine learning, and predictive analytics.

AI Case Study Blog Image

Supporting Strategy with AI

In this case, the organisation established high-level business objectives, such as "increase membership and minimise churn" and "ensure relevance of research, analysis tools, and customer events." Amplifi broke down these objectives to identify the necessary data capabilities:

  • High quality, available data, for customers and their interactions with the organisation research, analysis tools and customer events
  • A data organisation that provides monitoring, reporting, quality and governance services
  • Technology that supports advanced content management & search
  • Technology that supports analytics

An achievable value-add roadmap

Our priority was to understand the base data and put in measures and controls to ensure quality. By deploying Microsoft Azure as a data platform, we were able to bring data together for analysis while documenting it formally for the first time. Analysing and enabling data sets such as customer, publications, conferences, and consulting services, we built single views, data quality reports, and descriptive analytics that provided new insight. A data governance and quality operating model has been developed and is being deployed to continue the good work. These were the “hard-yards” but with the foundations in place business benefits can be delivered rapidly.

For example, trustworthy reporting on “who has interacted with which publication, for what reasons, and when” has enabled the predictive analytics use case of asking, “if we publish an article on a particular topic, how will it perform?”. We enabled the retrieval of the most semantically similar existing content to a novel research paper by splitting the organisation’s previously published research papers (and other material of interest) into 'text chunks' based upon logic (such as by paragraph, section, or subsection) and then converting these into vector embeddings to capture their semantic meaning.

The historical performance of the existing content is displayed through PowerBI reports and performance of the novel paper predicted. It is not an exact science, or a promise of success, but this data-driven indicator can inform investment decisions and directly support the outcome “ensure relevance of research”.

With the research semantically codified in this way, an opportunity for a GenAI use-case arose: a ChatGPT-style search engine drawing answers from the whole knowledge base.

Creating and deploying the LLM

In only 2 weeks, Amplifi delivered a PoC private Large Language Model (LLM) application that allows users with the correct credentials to ask the opinion of the organisation on any subject. The response is built from the statistically most relevant publications available to the user, with the publication and referenced sections provided for further reading. This can be deployed in a number of ways:

Internally, for use by account managers when they are asked a question by members

  • Enables efficient searching of unstructured content
  • Enables delivery of consistent messaging, even by more junior staff
  • Increases the value of the account management interaction with the member
  • Contributes to the outcome “minimise churn”

Externally, for use by members

  • Enables efficient searching of paywall restricted unstructured content by members
  • Increases the value of the web content interaction with the member
  • Contributes to the outcome “minimise churn

Externally, for use by the general public

  • Enables efficient searching of publicly unstructured content by non-members
  • Promotes paywall restricted unstructured content, encouraging non-members to sign up
  • Contributes to outcome “Increase membership”

All of these deployments make the “search history” data available. This data will be used to collect the trends and interests of the members to contribute to the outcome ensure relevance of research, analysis tools and customer events”.

The organisation will be deploying this internally in the coming months, with a view to rolling it out further. In the meantime, Amplifi will be identifying more strategically aligned use-cases and exploiting the robust data platform and processes.

This will ultimately result in improved data reliability, actionable insights through predictive analytics, and enhanced user interactions via a private LLM application. These improvements have already directly contributed to the strategic objectives of minimising churn, increasing membership, and ensuring the relevance of the organisation’s research and services.

3 key takeaways that can be learnt from this case study

1. Everybody “wants AI”, but the “want” should be aligned with the “needs”

While there is a strong desire across many organisations to implement AI solutions, it is essential that this enthusiasm is matched with the genuine needs of the business. This alignment ensures that AI initiatives are purposeful, effectively addressing real business challenges rather than being driven by a desire to keep up with trends.

2. Prioritise data governance and quality

Before the benefits of AI can be realised, a solid foundation of data governance, data quality, and data availability must be established. Understanding the current state of data and the effort required to improve it is crucial for securing funding and managing expectations. This involves recognising the 'hard-yards' necessary to ensure data integrity and reliability, which are critical for the success of AI projects.

3. Leverage robust data operations for rapid benefits

Once a strong data operation that aligns with strategic outcomes is in place, the organisation can swiftly reap the benefits of AI. There’s a high level of importance for investing in robust data infrastructure and ensuring that it is closely aligned with broader business objectives. Such an approach not only accelerates the realisation of AI benefits but also ensures they are sustainable and meaningful.

This organisation's case study is a brilliant example of how AI can be leveraged to achieve key business outcomes. The dynamic nature of AI means there are countless other possibilities waiting to be explored, unique to every organisation.

If you have an AI-based idea you want to turn into reality, read our guide for 6 expert tips for adding business value with AI. Not sure where to begin? Our data experts can help you determine your needs and devise a strategic approach, tailored to your organisation’s unique goals. Get in touch here.

Download Guide | 6 expert tips for driving business value with AI

Amplifi Driving Business Value with AI Guide Mockup