Last Thursday and Friday, I attended the Marketing Analytics Summit in London, an event that brought together experts and practitioners eager to learn about the latest in data-driven marketing.
Across two days, the sessions covered everything from AI’s role in creative ideation to the practicalities of device mining and navigating the “cookiepocalypse.”
While the event wasn’t my typical CRO/Experimentation setting, it was eye-opening to see the overlap with analytics and how both fields are evolving in sync. Between insightful discussions on data literacy, the future of attribution, and making customer data work smarter, the summit offered a wealth of ideas and practical takeaways for anyone invested in the power of data.
Here’s what I took away from each session in terms of actionable strategies and the industry shifts that stood out.
Unlocking AI’s Creative Potential with Jim Sterne
Jim Sterne challenged the usual narrative around AI as merely a tool for automating the mundane. Instead, he painted a bigger picture where AI is a springboard for creativity, helping us think beyond conventional boundaries.
He likened prompt engineering to a conversation with a stranger in a lively bar… at closing time. The conversation is entertaining, sometimes profound, but after a few drinks, you can’t take every word at face value. With AI the message was clear: engage deeply, check the facts, and push it to do better.
Jim’s “flip the script” approach to prompt engineering does exactly this. Rather than simply accepting an AI-generated response, he suggested prompting it to critique itself, scoring its answer from 1 to 10. If it scores below a 9, ask it to identify three questions that could improve the response. Then, have it try again, critiquing and iterating until it reaches a high standard. This method, Jim explained, isn’t just about getting things done efficiently: it’s about coaxing out more original ideas and richer insights.
Jim took things further, encouraging us to ask AI to go “off on a tangent” to explore different angles on a problem or even propose potential future scenarios and how to prepare for them. In his view, these creative prompts can lead AI to produce insights that might not surface through standard questions alone.
Looking to the future, Jim envisions AI agents becoming personalised, proactive, and deeply woven into our daily lives: anticipating our needs, curating options, and even negotiating on our behalf. For Jim, this isn’t far-fetched; it’s the natural evolution of AI, where its true potential lies in helping us ask better questions and think more creatively.
His parting advice? “Be creative with your questions.” In his view, the best questions lead to the most meaningful answers – and that’s where genuine growth happens.
Lina Mikolajczyk on Building Effective Data Teams
Lina’s session offered a refreshingly honest view on the challenges facing data teams today. She walked us through the “analytics maturity curve,” highlighting how many companies are stuck at basic reporting and self-service analytics, far from reaching true optimisation or monetisation.
For Lina, advancing data maturity is all about fostering trust, building partnerships, and ultimately creating tangible value – a journey she outlined in three essential steps.
Her first step, Trust, emphasised the importance of gathering candid feedback through surveys. Lina shared a memorable idea: running a bake sale in the office to encourage survey participation. Establishing trust, she noted, creates a foundation for the next step, Partnership, where data teams go beyond servicing requests to actively shaping business strategy.
The third step of creating Real Value is in part about always delivering a little more than requested – “give 10% more even when they don’t want it.” This practice doesn’t just add value to individual projects, it encourages stakeholders to engage more deeply and ask better questions, redefining the data team’s role within the organisation.
Her key takeaway was if data teams want to have a real impact, they need to rethink their own practices and create stronger, more collaborative connections with the business.
Doug Hall on Making the Most of Server-Side Data Collection
Doug’s session, “Beyond Server-Side Data Collection,” highlighted the value of keeping data in-house for greater control and privacy. By moving data collection off third-party tags and into their own cloud, companies can better govern and secure their data.
He explained how integrating first-party data with third-party sources, like location data, unlocks deeper insights for everything from real-time bidding optimisation to advanced attribution all while keeping data private.
A memorable part of Doug’s talk was his take on using AI to tackle bot traffic. Traditional CAPTCHA methods can frustrate users and pose accessibility challenges. In contrast, server-side bot detection keeps the experience seamless, assessing and scoring “bot-ness” invisibly without disrupting genuine users or creating barriers for those with accessibility needs.
Finally, Doug touched on the impact for marketers: server-side collection keeps campaign data clean and consistent. He wanted us all to understand that server-side data might seem technical, but the gains in privacy, accuracy, and flexibility make it a clear win for companies serious about data.
Neil Mason on Building a Really Offensive Data Strategy
Neil’s session focused on closing the “value gap” between business expectations and what data teams can realistically deliver. He explained how businesses often expect immediate and transformative insights from data, yet the complexity and scale of work required to deliver on these expectations can create a noticeable gap.
To bridge this gap, Neil introduced the concept of a “really offensive” data strategy. This is one that actively drives value rather than merely supporting business needs. A key part of this is reversing the traditional data flow: instead of starting with data collection, data teams should begin by identifying the decisions and outcomes the business seeks, then work backwards to determine what insights and data are truly needed.
At Creative CX, we’ve embraced a similar shift in perspective with our problem-first prioritisation approach. Just as Neil’s “really offensive” data strategy flips the usual flow (starting with desired business outcomes rather than raw data) our methodology prioritises customer problems before jumping to solutions. You can read more about that approach here.
Neil highlighted three main ways to close the value gap, data teams need to think offensively, focus on the last mile, design with users in mind, and build mechanisms that track outcomes of insights to ensure valuable data doesn’t get lost.
Juliana Jackson on Using Small Language Models to Close the Expectation Gap
While everyone else at the conference was buzzing about large language models, Juliana’s talk stood out by focusing on small language models. She shared how these models can offer a more secure and customisable solution.
Juliana explored the gap between customer expectations and their actions on websites. This gap often occurs when users don’t find the information they’re seeking or struggling to validate their choices. She identified three specific types of dissonance:
- Information-Related Dissonance: Users need more detailed information to make informed choices. Without it, they may feel uncertain and are likely to abandon the page.
- Navigational Dissonance: Poor site navigation frustrates users, making it difficult for them to locate key information.
- Decision-Related Dissonance: At key points in the purchase process, users second-guess their choices and need validation to feel confident in continuing.
To address these gaps, Juliana demonstrated how small language models, driven by natural language processing (NLP), can help connect search intent with page intent. This process involves three components:
- Premise: the underlying need or question in the user’s search.
- Hypothesis: an assumption about the information the user seeks.
- Classification: categorising the query to deliver responses aligned with the user’s intent.
For example, if a user searches “eco-friendly packaging,” the model could identify intent and highlight product details matching their values, increasing engagement. Small language models empower companies to craft relevant content while maintaining peace of mind surrounding data, an edge for industries where privacy is a must.
Phil Pearce on Reviewing 1,000 GTM Accounts
Phil’s talk looked into 1,000 Google Tag Manager (GTM) accounts, focusing on identifying common tagging issues and compliance risks. He used tools like Screaming Frog and Consent Mode Monitor to analyse GTM setups across various sites and reviewed patterns in tag management, version usage, and consent compliance.
His findings? Many sites had outdated configurations or incomplete consent setups that could impact their compliance with GDPR and other regulations.
Phil also shared some technical pointers, like handling bot detection during crawling and spotting inefficient tag configurations. The key thing to takeaway here is that regularly auditing GTM setups is important for compliance and performance, especially for high-traffic sites.
Martin Broadhurst on Generative AI in Digital Analytics
Martin’s talk on using generative AI in digital analytics was refreshingly practical and packed with actionable insights. He tackled the question on many minds (will AI replace analysts?) and provide a balanced perspective: AI is an invaluable assistant, not a replacement. Martin shared how AI can streamline tasks like generating ideas at scale, planning, and initial data exploration, making it a powerful tool when used thoughtfully.
Quoting Sam Altman’s prediction that AI could handle up to 95% of what marketers and strategists currently do, Martin reminded us that AI’s role is to support, not take over. He demonstrated some available tools, but also showed us that some integrated tools out there have room for improvement and can sometimes miss the mark on insights.
In a funny twist, just hours after Martin’s talk, Claude announced new advanced data tools which serves as proof that even the pros like Martin can’t always predict the next AI leap!
It was a perfect reminder of how quickly the AI landscape evolves. Chatting with Martin afterwards reinforced the message: use AI to boost efficiency, stay critical, and keep control over insights. A smart approach to AI could just make the role of an analyst more impactful than ever.
Craig Sullivan on Device Mining with GA4
Craig’s talk explored the often-overlooked process of device mining using GA4, a method for spotting issues across the many types of devices people used to access websites. Craig explained that many sites contain hidden bugs and usability issues that only surface when tested across a variety of devices, screen sizes, and browsers.
As he got us all to repeat out loud – “The device in my hand is not a proxy for what other users have.”
Optimising for just a few common configurations misses the varied experiences of real users. Craig shared striking examples where conversion rates dropped due to device-specific issues such as payment problems on certain mobile browsers or navigation failures on larger screens, demonstrating how ignoring these issues can quietly drain revenue.
He shared his Locker dashboard to pinpoint anomalies and device-related conversion issues within GA4 data. This tool helps teams save time by bringing device-specific performance issues to surface, so they can tackle bugs before customers encounter them.
Craig’s takeaway? Start by gathering the GA4 data to really understand how users engage with your site across different devices. Once you have that information, dig into it to uncover any issues or opportunities that might be hiding in plain sight. And finally, move on from simulators and emulators. Investing in real devices is essential for accurately assessing how your customers experience your site.
By committing to thorough device testing and letting the data guide your strategies, businesses can avoid revenue losses and create a smoother, more enjoyable experience for users on any device.
Scarlett Abraham on being A Problem-Solver Instead of a Data Person
Scarlett’s talk, ‘From Numbers to Revenue,” centred around her journey from an analyst to Head of Monetisation. She shared how adopting a problem-solving mindset over a strict data focus transformed her impact. Scarlett explained that moving beyond data for data’s sake helped her build stronger relationships, earn trust, and ultimately gain more freedom within her role.
She highlighted the importance of not solely identifying as a “data person,” but rather as a problem solver, which helped her navigate challenging projects. Her journey involved tackling issues with a practical mindset. Taking a broken Google Sheets process, for example, and building a sophisticated data-driven solution that solved core business problems without needing to overemphasise the data itself.
Build strong relationships, trust, and seeks to solve business problems first.
Scarlett’s approach led her to unexpected growth and even new career adventures, showing how stepping away from the purely technical side can open up your career possibilities.
Salman Alshuwaier on Building Data Literacy from the Ground Up
Salman Alshuwaier’s talk focused on the importance of data literacy as a foundational skill for any organisation looking to make informed, data-driven decisions.
Salman began by describing a data literacy survey that mapped the baseline knowledge of different departments, allowing him to tailor his approach for each team. For instance, while the CRM team had a strong data skills, other teams required more foundational training. To create excitement around data, he formed an Insights Ambassador Squad. This is a group of representatives from various departments tasked with supporting data initiatives and advocating for data-driven approaches within their teams.
He also introduced the Hypothesis Challenge Game to encourage hypothesis-driven thinking. By gamifying A/B testing, teams became more engaged in formulating and testing measurable ideas, making data feel accessible and fun.
Lastly, Data Day was a monthly gathering of ambassadors to share insights, introduce new hypotheses, and keep the momentum around data literacy alive.
Salmon’s approach underscored that building data literacy isn’t about pushing numbers but about fostering a shared enthusiasm for data that ultimately supports better business decisions. A rising tide lifts all boats!
Robert Petković on Consent Mode: The Pitfalls and Best Practices
Robert’s session, “Consent Mode: So Easy… to Mess Up,” took a long look into the complexities of consent management and compliance, especially in the EU’s regulatory landscape. He shared common mistakes in consent setups, from tracking users before obtaining consent to placing banners in hard-to-find spots, all of which can impact both user trust and data quality.
One of his key points was that good consent implementation isn’t just a legal formality; it’s essential for quality data collection and effective marketing. Robert highlighted the importance of selecting the right consent mode: Basic mode blocks tracking until users consent, while Advanced mode can collect data anonymously if users opt out, allowing for behavioural modelling without compromising compliance.
I liked how Robert illustrated successful practices by referencing real-world examples like Sephora’s shopping basket system and demonstrated practical tools, such as Cookiebot, to streamline consent processes.
His advice could be summarised to get data, legal, and marketing teams to work together to ensure consent isn’t just implemented correctly but also enhances data reliability.
Yael Farkas on Combining Attribution, MMM, and Incrementality
Yael’s talk focused on how Douglas implemented Marketing Mix Modeling (MMM) to enhance marleting efficiency across its operations in 22 European countries. Faced with challenges like data silos and the complexity of multiple markets, Yael adopted a 360-degree approach by triangulating data from MMM, incrementality testing, and multi-touch attribution. This helped Douglas accurately assess each channel’s contribution and break down silos, making data more accessible across teams.
Yael also stressed the importance of a well-structured data and tagging strategy – “If you want to claim your results, you have to put in the work to track them properly.” Her team’s efforts evolved from a data project to a tech and people project, requiring cross-functional collaboration to be successful.
Ian on Navigating the Cookie Apocalypse with AI
Ian’s talk, titled “Can AI Save us from the Cookie Apocalypse?” tacked the pressing challenge of third-party cookie deprecation and the changing landscape of digital advertising. With the impending “cookiepocalypse,” Ian argued that the reliance on third-party cookies is outdated, as privacy concerns continue to reshape advertising strategies. Google’s Privacy Sandbox aims to address this, but he cautioned it may consolidate even more power within Google’s ecosystem which has lead to some regulatory concerns.
He outlined three AI-driven strategies to adapt in this new environment:
- Semantic Contextual Targeting: Moving beyond basic keyword targeting, this approach leverages AI to understand the deeper context of content, helping ads align more naturally with the environment they’re displayed in.
- Creative Automation: By using AI to create and iterate on ad creatives at scale, marketers can ensure variety and freshness, reducing ad fatigue and keeping campaigns relevant.
- Continuous Optimisation: Ian discussed an always-on approach where AI continuously tests and optimises creative variations, ensuring ads remain engaging and reducing the typical performance drop-off seen with static campaigns.
Ian’s message was that its time for marketers to rethink their dependence on audience targeting through cookies and embrace AI-powered strategies that offer privacy-conscious, contextually relevant, and creatively flexible advertising solutions.
Conclusion
Looking back, the summit was as informative as it was welcoming. Analytics and CRO share a lot of common ground, and this summit really showed how much both fields can gain by learning from what the talking points are in both industries. I left with new strategies, tools, tips and plenty of inspiration. And with the collaborative spirit in the room, it felt less like just another event and more like finding a community of peers.
