What we've learnt building an AI accuracy checker

It's time for another emergency episode!OpenAI has announced the launch of ChatGPT Business Hub, a tool that promises to give businesses a place to control the information ChatGPT has about them, see how they perform, and interact with customers via an agent.Given our theme of discussing 'signal loss' this quarter, could this be a significant gift to marketers crying out for better performance data?Chapters00:00 Introduction and Context02:50 Overview of ChatGPT Business Hub05:51 Implications for Businesses08:27 Advertising and Marketing Insights11:27 User Experience and Setup14:19 Potential Impact on Financial Services17:10 Future of AI Tools and Competitors19:50 Conclusion and Jargon Buster

With an impending Competition & Markets Authority deadline looming, is Google about to unleash a better set of AI search stats for website owners?Right now, marketers and publishers can see how many impressions they get in AI search via Google Search Console.An upcoming regulatory deadline in the UK wants the search giant to go further, so could we be about to see a rare improvement in the information available?Chapters00:00 Meet the panel and introduce the signal-loss theme01:15 The CMA’s concerns: choice, transparency, and attribution04:06 What Google Search Console currently reports06:34 Which Google AI features could the rules cover?09:22 Why marketers want AI search query data12:15 Using search queries to spot audience and content gaps14:57 Will Google add click reports by the December deadline?17:00 AI Mode monetization and Google’s incentive to withhold data19:16 How advertising and search measurement may change26:28 What compliance could look like for publishers29:36 Is it fair for marketers to expect the data?31:21 The wider commercial context and the December milestone33:36 Jargon buster: what EBITDA stands for

It's launch day for our new product, Corrik, so Pat and Paul talk through what they've learned in building and testing it.Corrik is an AI visibility and accuracy checker. It uses a 'source of truth' to flag when LLMs are quoting inaccurate information.The team have spent the week testing it in the financial services sector and have uncovered where it's strong, and where it needs work.If you're interested in the challenge of tracking AI accuracy, this is one for you!Chapters00:16 The PHP Vulnerability Crisis07:15 Introducing Corrik: An AI Accuracy Checker15:32 Testing Corrik: Insights and Findings26:29 Managing Online Information Accuracy31:31 Corrik's Launch and Future Plans36:11 The Importance of a Source of Truth40:55 Jargon busters!

Last week, FCA CEO, Nikhil Rathi appealed to attendees of the FCA's Annual Public Meeting to continue to come up with inventive ways to track consumer outcomes.With the challenge fresh in mind, we went away and built a tool that could potentially help do just that.Using a database of around 90 search terms, and industry benchmark, and Google Trends, our Consumer Duty Search Index looks at how consumers are using search to support them.Today, we dive into the key findings!Chapters00:00 Introduction and episode overview02:33 Applying Google Trends to financial services and consumer duty03:56 What is Google Trends and how does it work?06:08 Using Google Trends to track flu symptoms and outbreaks10:12 Building a search term database for consumer outcomes13:15 Key findings on search trends in finance15:21 Case studies: market reactions to events16:26 Search trends related to vulnerability and scams16:52 Impact of news cycles on search behavior19:58 Using search data for regulatory insights and firm strategy28:58 How firms can leverage search trend data37:29 The role of AI and LLMs in consumer search behavior38:44 Limitations and future potential of search data40:41 Conclusion and next steps for firms and regulators

This week saw the FCA host it's annual public meeting in Edinburgh.Hosting over 900 attendees both online and in the room, the FCA were quizzed on topics from how it works with whistleblowers, Consumer Duty, AI, and more.We pick up on the technology and marketing angles present in the event and unpack what we took away.We look specifically how the FCA does (or doesn't) oversee AI, and what that means for firms in the industry.Chapters00:00 Introduction to the FCA Public Meeting and Key Themes03:25 Human Element in Financial Regulation05:28 Handling of Criticism and FCA's Response to Whistleblower Questions11:14 AI Developments and FCA's Upcoming Papers13:11 Regulation of AI and Consumer Duty Framework14:47 Financial Advice from Large Language Models16:12 Firm Accountability for AI and Technology Use18:05 Targeted Support Regime and Advice Gap20:08 Regulation of AI and Fast-Moving Technology21:58 Lead Generation Crackdown and Data Misuse23:50 Enforcement Actions and Data Provenance25:59 Data Gathering and Observational Data Use27:42 Consumer Duty and Market Impact31:55 Vulnerable Customers and Power of Attorney Issues37:21 Accessibility challenges and power of attorney38:23 The FCA is appealing for support in assessing Consumer Duty outcomes39:56 Using Data and Trends to Detect Harm42:44 Search Trends and Financial Harm Indicators43:32 Jargon Busters: Fiduciary Duty

It's launch week for Q4 and this edition is all about Signal Loss.Whilst demand for accurate reporting data is as high as it has ever been, access to clean and reliable web data is arguably worse than it has been since the launch of the web.To kick things off, we dig into cookie compliance and why Google Analytics (or GA4) isn't anywhere near as accurate as it used to be.Thankfully, we have a few thoughts on what marketing teams can do ahead of planning their 2027 marketing strategies!Chapters00:00 Introduction and episode overview00:16 Meet the hosts and episode context01:03 Series focus on trust and signal loss in Q402:12 Setting the scene for tracking challenges in 202605:23 Technical overview of web tracking and cookies07:18 The decline of Google Analytics and cookie compliance12:03 The rise of third-party tracking and privacy laws17:30 Browser controls and their impact on tracking20:46 GA4 and the limitations of cookie-based tracking23:23 Impact of cookie banners and user opt-in behavior26:12 Data segmentation and bias in tracking data30:42 Triangulating data sources for accurate insights31:41 Design considerations for cookie banners and UX35:39 The future of cookies and alternative tracking methods39:55 Google’s evolving approach to privacy and ad targeting44:48 Emerging AI tools and privacy concerns46:26 Key takeaways: triangulation and data reliability47:05 Jargon busting: understanding junk bonds

In July (2026), Cloudflare announced that it was going to categorise web crawlers (bots) in one of three ways: search, agent, or training.If you want to opt out of training, you can block training bots - easy, right?Not so...common crawlers such as Googlebot (critical for showing up in Google search) are classed as multi-purpose. They crawl for search purposes, but they also train AI models.In this discussion, we explain Cloudflare's strange decision to hide settings within the 'security' section and how easily a brand could trap itself into obscurity if it chooses the incorrect settings.Chapeters00:00 Introduction to AI training and web crawling issues01:58 Overview of Cloudflare's categorization of crawlers03:31 Cloudflare's default settings for training and agent crawlers04:27 The challenge of 'multi-purpose crawlers'06:19 What is Cloudflare and how does it work?08:14 The evolution of Cloudflare's features and internet security09:28 The rise of AI crawlers and their impact on content security11:36 The challenge of balancing bot access with website security15:31 Implications for businesses and website owners21:32 Managing crawler settings and the role of IT and marketing24:25 Historical context and the evolution of AI and web scraping25:48 The decoupling of impressions and clicks due to AI overviews28:17 Cloudflare hides crawler settings within 'security'29:39 Next steps for businesses: managing Cloudflare settings33:20 The importance of transparency and security in bot management38:30 The future of website visibility and AI training42:02 Jargon busters and closing thoughts

Fresh from our annual security training with Pat, the team discussed AI-driven fraud in our industry.It's a growing challenge; according to Mitek/Datos, U.S. unsecured credit losses reached ~$2.94 billion in 2025, up from $1.8 billion in 2020.We cover some horror stories, like the UAE banking exec who was scammed by a 'deepfake' of a company Director into initiating $35m in transfers.But AI is also at the forefront of the fight against fraud. Banks are flagging potential issues sooner and apps are using smart features to show when you are and aren't on a legitimate call.Chapters00:00 Introduction to AI's impact on fraud01:48 Statistics on AI-driven fraud increase03:44 Deepfake scams: case studies from 2020 and 202407:09 How scammers use AI to craft convincing emails and calls12:50 The role of AI in enhancing verification and security13:37 Who owns the role of tackling AI-driven fraud?18:21 The evolution of scam sophistication and AI's role22:29 The concept of AI prompt injection29:58 Using AI in the battle against fraud31:31 Balancing security with user experience33:40 Invisible security measures and behavioral signals36:52 Balancing data sharing for security versus privacy41:39 Innovative UX features to prevent scams46:47 Future outlook: AI in security and fraud detection49:04 Jargon busters: amortisation

This week, Paul and Amelia sit down for a chat with Nick Laws, the senior creative leader behind work with Nest Pensions and the recently rebranded Centre for Inclusive Money (formerly Nest Insight). Nick chats about the tricky task of engaging consumers with their pensions. We touch upon his work alongside Aardman (of Wallace & Gromit fame) on building a national ad campaign. We also discuss the role of influencers in financial campaigns; including the Everyday Investor campaign featuring a former winner of The Great British Bake Off who baked using products that benefited from Nest pension pot investment. It's a fascinating discussion covering one of the most lively parts of the world of financial services. Chapters 00:00 Introduction to Nick Laws and his role at Inclusive Money02:28 The challenge of engaging consumers with pensions04:17 Complexities of pension types and communication hurdles06:32 Creative restrictions in financial promotion09:28 Working with Aardman on the Nest brand awareness15:37 Auto-enrollment and consumer inertia in pensions21:43 The Everyday Investor campaign and innovative engagement30:24 Impact of generative AI on creative work37:54 Working on a mission-led brand and its unique aspects39:36 The importance of purpose and authenticity in branding40:59 Jargon busting: Understanding the black swan event

We've just updated our big guide to SEO in financial services, so this week the marketing team got together to define what SEO actually means in our industry.We outline the challenges banking, insurance, wealth management, and fintech firms face in search and what good search strategy looks like.We also outline our case for why search is a relevant factor when it comes to Consumer Duty.Chapters00:00 Introduction to SEO in Financial Services01:09 What is SEO and why is it different in finance?04:36 Unique challenges of finance SEO: YMYL05:49 EEAT: Experience, Expertise, Authority, Trustworthiness07:01 Regulatory and compliance considerations in finance SEO11:40 Subsector differences: Banking, Wealth Management, Insurance, Fintech16:38 SEO in wealth management19:55 SEO for insurance brands22:03 SEO for fintechs23:22 Content strategies: FAQs, keyword research, and outdated content27:12 Powerful keyword research using a brand name29:28 Building a strategic approach: Technical, Content, Promotion33:24 Navigational search and user journey optimisation36:40 Impact of AI and generative search on SEO in finance42:28 Consumer duty and the importance of accurate search results45:59 Why SEO is a Consumer Duty factor46:37 Jargon Busters: AUM