top of page
  • Writer's pictureICS.AI

New College Lanarkshire share how they are transforming student retention with AI

Updated: Sep 16, 2022

We were recently joined by one of our founding Further Education (FE) customers - New College Lanarkshire – at a webinar to talk about their success with Artificial Intelligence (AI) and Machine Learning in transforming student recruitment and retention (watch the recording here).

This ground-breaking project will enable New College Lanarkshire to fine-tune resource distribution and provide 24/7 digital support quicker. We explored what it means for NCLan students with Allan Forsyth (Head of Information Systems and Development) and Matthew Smith (Chief Transformation Officer). We also touched upon their successes to date, the opportunities ahead for the whole education sector, and why they chose to work with us.

Here are some of the talking points from our Q&A interview.

How did we start working together?

We’ve been interested in using artificial intelligence (AI) to analyse student behaviour for a while. After attending the ICS.AI webinar detailing the success Durham University have had with their AI Assistant implementation, we were impressed by the technology and how quickly it was able to provide an effective solution. You helped us develop our ambition of using AI to look at the data and understand how it might help improve our services.

Machine learning is a new technique in this space to analyse student data. What techniques had you tried previously?

We had only dabbled in our review. Another college had done quite a lot of work on this themselves a few years back; we were very impressed by that and wanted to do something similar. However, we concluded we didn't have the resources or expertise to be able to implement it ourselves. Even if we developed the expertise, we didn't have the kind of strength and depth required to implement it effectively. When we started talking to you, we were at that point where we needed help. The timing was perfect, as the technology had matured to a level that was ready for exploitation.

That’s a great point! This is an evolving technology and we can see it improve customer to customer. The rate of development and number of iterations we could go through with you means this project is ground-breaking. We now have great insights into what predicts different behaviours – but, when we looked at the final model, we were surprised by how strong the correlation was between recruitment and retention of students. What are your thoughts on that?

We weren’t surprised that there was a correlation. A lot of those factors were ones that followed conversations we’ve had within our team. What surprised us was how strong and direct the correlation was. The strongest one was the point of application, where in certain scenarios there is a straight-line decline. Most of our courses start in August - if students apply for the course in January, they’re very likely to stay. For someone that applies in July and August there's a high chance they're going to withdraw. To see that from the analysis changes the conversation internally – when you’ve got that direct evidence then you don’t have any excuse other than to take action!

The final machine learning model we developed was complex, but we were able to simplify it and produce a straightforward calculator for you. How useful is that going to be for the college going forward?

We were able to take a straightforward formula, apply it to our existing data and make predictions without doing anything too complex. It means we can now apply that formula going forward - we are planning to target interventions based on that. For example, we have a team at the College called ‘Key Support’. Their job is to take referrals from teaching staff for people who are considered to be at risk (perhaps because of their attendance). Historically, they would contact the student and see what support they might need. The success of this to date is variable, as it’s all dependent on the teaching staff. Now, using this formula, we will be able to supplement that system; targeting and intervening with at risk students earlier i.e. when they start to not attend. We capture that information in this first, learn from it and see which interventions are useful and which are not as effective.

To be effective, those interventions can be time-consuming and require a high level of input from staff. With the well-publicised staffing challenges in the sector, how do you think the digital assistant we’re now working on is going to help with freeing up staff-time?

We hope it will help us in a big way. Together with ICS.AI, we're setting up a digital assistant (chatbot) that will deflect the more routine enquiries, especially at peak crucial times i.e., just before a course starts. Staff are always under pressure, but at these peak times it's exceptional. If we can reduce the amount of time they are spending acting as signposts and pointing people to existing resources, it's a huge benefit. We want our staff to not only be helping the students who have more complex queries, but also to spend more time with them and provide the support and reassurance that only the staff can provide.

How do you see engaging with your students in this way beneficial in the long-term?

It will provide a better service. It’s always been the case that students live different hours from the working population. Online enrolments and online learning activity is happening all hours of the day. The digital assistant will be there 24/7, which our staff can’t be. That alone will make a big difference. It can always provide a consistent answer; students come to the College from different paths, and they will approach different departments. The digital assistant will ensure consistency.

What does the leadership team think of the project to date?

Our Executive Board are excited about the opportunity, the power and potential of the AI project. They are supportive as they see the direct benefits to our students and to our retention; both staying on the programmes and being successful at the end of the programmes. We’ve presented the initial findings to them, which were very powerful. They are looking forward to seeing the success of the project. Overall, we very much look forward to the next academic year, continuing to roll-out our Machine Learning model and digital assistant and seeing the benefits in terms of early retention.

A final question - how do you think this sort of approach is going to help the education sector in general?

AI systems are being used more and more. Because of the always-on 24/7 nature of the chatbot I see big opportunities for transformation. We’re very excited about it and look forward to seeing the benefits unfold further as the year goes on.

NOTE: This interview has been extracted, transcribed and edited from a live webinar event.

Summary solution text:

Working with New College, we have successfully used machine learning to produce highly accurate models from a combination of structured and unstructured data. Whilst it is well known that key factors such as attendance rates affect performance, modelling the complex interplay of multiple risk factors we have allowed New College to identify a small cohort of students who have a significantly higher risk of failure early in the academic year. We have then worked with the college to develop specific AI tools to free up staff time and allow delivery of targeted interventions that not only improve student performance but also student retention. Our AI assistant further allows the college to engage with students to get help, raise queries and book appointments further driving student satisfaction and retention.

To see demos of this product - Watch the webinar!


ICS.AI are the UK's market leader in conversational AI solutions across the public sector. Based in Hampshire, we have customers across central and local governments, higher education and healthcare. As a Microsoft Gold Partner we have delivered channel shift and digital transformation with our proprietary SMART range of AI products.

8 views0 comments


bottom of page