In September 2026, the Experience and Insights team travelled to Big Data LDN. For decades, Appius’ data scientists have been helping organisations manage huge volumes of customer data, integrate systems efficiently and embed structures into new and existing solutions for accurate tracking. Today, data science spans across everything we do at Appius, from marketing to front end development. It has never been more important for our team to stay up-to-date with the latest trends.
Technology is moving so quickly, and accurate data is becoming vital as companies integrate more advanced AI systems into their already complex tech stacks.
Whilst AI is by no means a new subject, events like Big Data LDN ensure Appius are at the forefront of what’s happening today, and what’s likely to happen in the future. Most importantly, we discover how to take all this new technology and turn it into something useful for our clients and the people running marketing and technology departments behind the scenes.
Here are the five biggest trends worth taking note of if your role has any crossover with analytics, big data and insights:
1. Small wins can lead to bigger changes
Kathleen MacLean, Global Head of Customer Grown Strategy & Planning at Dr. Martens, presented a fantastic example of managing customer data at an enterprise level. The team stopped prioritising long projects and started targeting small wins using proprietary data for decision making.
Using Amperity, Dr. Martens uploaded proprietary customer audience data into its paid marketing channels, allowing campaigns to target people using real customer insight rather than relying solely on platform data. This relatively small change delivered a triple-digit uplift in conversion, helping prove the commercial value of using first-party customer data to improve audience targeting.
With this transformation strategy, Dr. Martens tripled digital uplift into conversion from paid marketing channels. But success isn’t just measured in sales. MacLean introduced a three tier framework to prove success:
Increased engagement
The new customers discovering Dr. Martens were spending more time, taking more action and less likely to leave the page without engaging than they were before they added customer data into the audience targeting
Purchase intent
Ecommerce conversion journeys might end with the sale, but they start with product exploration and move to add to cart events before any transaction takes place. Kathleen MacLean used these behavioural signals to prove intent improved after data was added.
Purchase
The one metric stakeholders care about, how much money was driven by this small change.
If you work in data, you’re used to complex challenges and stakeholder opinions. But this Dr. Martens example shows you exactly what you can achieve when you break it down into specific problems. By recognising the issue, you can identify what success looks like giving you a benchmark to measure and test.
For marketers, this could be something as simple as campaign audience targeting. Make one small change to a targeting field, or upload proprietary customer data, then test and learn as you go.
Too many changes at once can muddy the insight. Focus on small, manageable changes that can be measured and attributed to results.
Learn what works.
Do it again.
2. AI needs context, not just more data
Having lots of data isn’t enough. Almost every speaker at Big Data London 2026 highlighted the importance of quality, accuracy and context. Before AI can magically answer a question or make useful decisions, it needs to understand what each piece of data means to an organisation.
One example really stood out:
An AI agent was asked to identify “premium customers” without any additional context. The agent sourced a list of the top customers based on lifetime spend. However, once the organisation supplied its definition of what a premiums customer looked like (acquired through email or search, aged 30+, at least three completed orders), the AI agent produced an entirely different answer.
To implement AI systems effectively and drive business growth, organisations must add a semantic layer. Think of this as a dictionary, explaining in plain-English language what things like revenue, customer, conversion or premium customer actually mean.
In an example case study, one organisation had implicitly stated that revenue is defines as sales from completed orders that had not been returned. And it’s not just definitions. A semantic layer can hold calculations, exclusions and relationships to ensure different teams are working from the same source of truth. AI doesn’t remove the need for good data governance. It makes it more important.
Before you ask whether your data is AI-ready, make sure your organisation agrees on what the data actually means.
3. Getting answers from data is becoming more accessible
LLMs are making everyday tasks easier, especially when it comes to data analysis. But whether we like it or not, ChatGPT and Claude can’t replace human expertise (yet). With that being said, Compare the Market have been utilising conversational analytics across their executive and data teams to build more efficient reporting processes.
Emanuela Ciotti, Head of Enterprise Analytics at Compare the Market, explained how the company has moved away from dashboard reporting entirely. As any data scientist knows, dashboards can be rolled out endless. But even if they’re dynamic and clever, users will still ask the data team for the answer.
Unless they can use conversational analysis.
With the help of Aman Bhatterai, Head of Enterprise Customer Outcomes at Thoughtspot, Emanuela explained that customers trusted Compare the Market more than ChatGPT with their personal and financial data. This led to a complete transformation of the way the team worked.
Instead of asking the data team and waiting for the answer, the team can get a quick response from the conversational analysis platform. Easier access can help more people use insight, provided the underlying information and definitions are dependable.
Even at Appius, our SEO team have implemented the Semrush MCP for conversational data analysis. It’s enabling faster causation connections and more insightful next steps for SEO and GEO strategies. We’re also seeing Google roll out new AI features in GA4, Google Search Console and Google Ads, enabling conversational analytics across the three platforms.
This new way of working is already being adopted across various business disciplines. If you’re not already thinking about it, it’s about time you started.
4. Good automation knows when people need to step in
AI has streamlined automation. It’s easier than ever to reach customers at the right time with custom messaging that supports their journey. Adding automation into the mix has also been proven to improve customer service response times and reduce required resources.
But easier automation comes with its own set of challenges. We’ve all heard the stories of AI customer service chatbots going against company policy.
Pratyusha Singaraju, Senior Software Engineer at Netflix, explained how they’d handled that exact issue with content workflows that combine predefined rules, machine-learning models and human decisions.
Netflix have stripped it’s workflows back, segmenting those that can be managed with rules, those that need machine learning and data model and those that need human involvement before they are actioned.
These workflows control what tv shows and films are being recommended to you. Some decisions are fully automated by rules, others use machine learning models to understand each users behaviour and intent. What’s interesting is the human involvement layer, even something as simple as a film recommendation comes with risk when you let AI take the reins. Netflix’s human layer steps in for the processes that they are really needed, AI and automation handles the rest.
The most important thing to take away from Netflix’s experience is simple. AI systems can be effective in real consumer facing platforms, but not everything should be totally automated. The human layer allows efficiencies in the right places with the right level of risk.
5. Trust needs to be part of the brief
Jamie Petherbridge-Conroy, Senior Principal Software Engineer at the BBC, highlighted some of the challenges they face compared to more commercial competitors. The end goal is to deliver personalised experience across platforms like BBC iPlayer in an inclusive and fair way in line with editorial standards.
Building on what Netflix had spoken about earlier, BBC added a human intervention layer with reliable ‘in concert’ machine learning. By embedding key principles, such as shielding vulnerable users, into the processes, the BBC maintained consumer trust whilst delivering personalised experiences that aligned with the organisations core values.
But in the closing keynote, Louis Theroux reminded us that storytelling trumps the truth in terms of human motivations. Good storytelling is what gains an audience and builds respect, where truth and cold-hard facts can be too boring.
When filming, Theroux admits to having a moral agenda. But he hides it to ask the factual questions and allow the interviewee and the subject matter to expose itself for what it is.
That is exactly what we’re dealing with when it comes to AI in 2026. Information and morality are not connected, until the humans make the connections. Trust can be broken almost instantly when false information is supplied. That’s why organisations implementing digital and AI systems into their processes need to train the ML models to question the supplied data’s morality.
To see true growth as we prepare for 2026, customer trust, accuracy and usefulness should be embedded in the definition of success.
Big Data LDN 2026: Final Thoughts
And that’s a wrap (for another year at least).
It’s never been more important to connect the people, platforms and processes that matter to your commercial objectives. Now we’re introducing AI into more of these processes, quality data and secure platforms enable growth.
If you feel like you’re falling behind, you’re not alone. We work with enterprise organisations every day to transform their stack, modernise the technology and build the data layers that make a difference. Reach out to our data team if you’d like to run through some of your challenges and find out how we can help.
You might also like
