CASE STUDY
Looking back, was there a first use case for AI that you can talk about?
A lot of what we call AI is actually machine learning and machine learning has been around for a very long time. VisitBritain was using machine learning even before I joined in March 2020.
However, the first time I saw it being truly embraced with real gusto was in our data harmonisation and master data management. A lot of the data we get from tourism is unstructured or semi-structured, so data governance and master data management are incredibly important. One of our first major uses was to make that process easier and less manual, while still delivering better outcomes. This means our team spends less time on data preparation, which is still vital because the human element matters and more time on high-value tasks: getting insights, telling stories, and putting information into the hands of those who need it.
Our first significant piece of work applying this was with our“ brand tracker” data source. This is a large survey we use to evaluate campaign performance. It typically contains tens of thousands of free text responses, which are impossible for a team of our size to analyse regularly with consistent quality, as human fatigue can set in. We now use Large Language Models( LLMs) to quickly qualify the data within seconds – identifying what’ s suitable for analysis and what isn’ t, perhaps due to blank or incoherent responses.
From this clean data, we can identify key topics and sentiments, creating summaries tailored to different needs. Our analytical marketing team can delve into detailed insights about culture, locations, or creative elements to inform new campaigns.
Marketing directors or senior staff can quickly grasp the top three takeaways for campaign improvement. And for the C-suite, we can provide an“ elevator pitch” – three to four lines summarising the campaign’ s impact. This was never possible before, certainly not within minutes, which is now the time it takes from data landing to getting these valuable outputs.
Have there been any big surprises that have resulted from those insights, where you’ ve pivoted on a campaign because you’ ve received an outcome you weren’ t expecting?
What is surprising, however, is some of the minutiae. For things that are working well, we now get to really understand the‘ why’ behind their success. Similarly, if something isn’ t being received as well – perhaps a particular creative was seen as too clichéd, or something shown didn’ t resonate for whatever reason – that information now comes to the forefront much sooner than ever before. This means when we go into planning conversations about creative briefs, it’ s a much richer discussion. That specific output is now taking centre stage, rather than just being another data source we glance at. The whole dynamic of those briefing and planning sessions has fundamentally changed.
What success indicators has VisitBritain seen particularly from its AI usage, and which trends and campaigns are standing out to you more now that you are using this technology?
I think the main impact is probably more internal than external at this stage, largely because, as I mentioned, the purchase cycle in tourism means direct external results take time.
However, if I pick on some key internal shifts, the first major difference is the point at which analytical professionals are brought into a conversation. It’ s now happening much, much sooner and given far more prominence. We already had a culture of data-first, data-driven decision-making, but that’ s now gone up several levels.
Secondly, the range of people in these conversations has expanded dramatically. Traditionally, you’ d expect individuals from my background to lead these discussions. Now, I have people in marketing, HR and other departments talking about data and AI with genuine authority, which is fantastic. The nature of the questions being asked has also changed; people are becoming more demanding, which I welcome because it pushes us further.
Where it’ s truly making a difference is in ideation. We now have sandboxes where non-technical people can collaborate with technical colleagues to experiment, combining different data sources. This allows us to explore how we can evaluate marketing more effectively. Furthermore, the clean data that is a byproduct of our AI structure has an invaluable secondary benefit: it unlocks other uses.
No, not really. We have a very good pulse on things, and our brand tracker is an incredibly powerful tool. The methodologies we use to evaluate campaigns, coupled with our integrated decision-making process, mean there aren’ t many major surprises.
Our ability to do marketing analytics has completely transformed. We now have access to much cleaner, more granular and structured data in a platform that enables us to do significantly more. Instead of just stating how campaigns performed based on basic metrics, we
www. intelligentcio. com INTELLIGENTCIO EUROPE 15