Attribution and data accuracy have always been at the heart of effective digital marketing, and in 2025, they’re more complex than ever. With the rise in VPN usage, Google’s AI Overviews reducing traditional site clicks and searchers turning to LLMs for answers, marketers are facing a major shift in how user behaviour is tracked and understood.
Combined with ongoing challenges around cookie consent and data privacy, the digital landscape is forcing advertisers to rethink how they measure performance and prove impact. To unpack these changes, we spoke to Loom’s Senior Digital Strategist, Lukas, about the current state of data, attribution and what businesses can do to stay ahead.
Jump straight to Lukas’ thoughts on…
VPN usage
AI Overviews
Searches on LLMs
Cookie changes
What’s next for measurement?
VPN usage
Why are we seeing a rise in VPN usage?
The introduction of the online safety act in 2025 led to websites which host adult-only content requiring verification from UK users that they are old enough to view this content. As this requirement doesn’t exist in many other countries, UK users have been utilising VPNs to change their location while browsing and bypass age verification requirements.
What is the impact of this on digital advertisers when it comes to tracking?
VPN can impact tracking in different ways:
- Many VPNs come with built-in anti-tracking features that block tracking scripts, or prevent cookies being set.
- VPNs lead to inaccurate data on a user’s location
They can also impact how we can serve ads. Many VPNs come with built-in ad-blocking capabilities. Since advertisers will usually target users in a certain country or countries, if users are showing as if they’re no longer in the UK, then it impacts how many people in the UK can accurately be targeted. This can be somewhat circumvented by focusing targeting on placement (e.g. on a UK-based newsite) or using user-provided data about their location shown in the platform (e.g. Linkedin or Facebook). On the flip side, advertisers targeting regions outside of the UK may show ads to people in the UK who are using a VPN in their country, that may often not be relevant for the UK user.
Do you see this as a big problem or something we can work around strategically?
Web analytics and ad platform data has long been something where we’ve had to work with non-perfect data. Between challenges around cookie consent, ad blockers and browser tracking prevention technology, marketers have always had challenges getting accurate data.
This means we’re often just looking at a sample of the data. We can account for this by estimating the impact- for example, by comparing the number of organic clicks we see in GSC against the number of sessions from google organic search in the same period.
AI Overviews
With AI Overviews resulting in reduced clicks to sites, what does that mean for traditional attribution models?
As users use AI Overviews more, particularly in the top of the funnel, we lose some visibility over our top-of-funnel impact, making it harder to see what role our organic marketing efforts are having. Users might gain awareness of our brand from seeing it in an AI Overview without clicking but then we can’t connect that to a sale in a later session.
How can businesses track their visibility/results when actions are being taken in the search results rather than on their own sites?
Google search console reports impressions for AI Overview results but currently doesn’t split these out from regular SERP impressions, which limits the value of this data.
Many brands have reached towards methods like incrementality testing and marketing mix modelling which aim to show how marketing efforts benefit sales overall rather than attempting to attribute every conversion as a response to worsening visibility of marketing efforts and outcomes.
Are impressions or engagement metrics enough to measure value?
They can be useful indicators but there are also other methods, such as using first-party data or incrementality testing and marketing mix modelling.
Searches on LLMS
When users search on LLMs instead of traditional search engines, what tracking signals can businesses rely on?
We can use web analytics platforms to see the traffic from AI Overviews and LLMs to our websites but these often don’t paint the full picture of all the visibility our brands are having in LLMs. There’s been an influx of new tools which can show brands how often they are showing up in LLM responses, this includes both new companies and existing SEO reporting tools like AWR and SEMrush.
Are there any industries that will be hit hardest by these changes in search behaviour?
Informational, news and entertainment sites that rely on ad revenue are hit the hardest with users no longer visiting sites.
This change will put a significant pressure on certain organisations to change their business models. Companies offering products and services which take more consideration, particularly in the B2B space are also more likely to be significantly impacted as they often rely on making users aware of their services through informational content and then remarketing to them, something they’re not able to do if a user reads their content via an LLM or AI Overview.
how can a business track SENTIMENT and use it?
Several LLM monitoring tools offer sentiment tracking. Finding out what LLMs are saying about your brand, service or product can allow you to tailor your website’s messaging and your advertising to ensure that users are seeing what you want them to see.
Unlike website messaging, you’re not completely in control of this, as online spaces like reviews and social media platforms are taken into account too. This has led to some brands developing more of an organic presence on social media platforms that they might not have previously, such as Reddit which has been cited as one of the most common sources of LLM responses.
Cookie changes
Where are we with Google’s cookie depreciation?
Since 2020, Google has repeatedly suggested it would limit how cookies are used in Google Chrome. This was repeatedly pushed back until early 2024 when the change was implemented for 1% of Chrome users.
But by July 2024 Google seemingly U-turned after intervention from the UK’s Competition and Markets Authority which felt that the replacement technology, Google’s privacy sandbox, would give Google too much power. Google is now working with the CMA and companies in the advertising industry to find a new solution.
How does this affect the accuracy of data?
Whilst Google Chrome hasn’t removed 3rd party cookies, Safari’s ITP has impacted how long 3rd-party cookies can last (as little as 1 day) and how they behave between different sites. This has made it harder to connect ad interactions to conversions and multiple sessions for the same user.
What alternatives should businesses track?
e.g. first-party data, server-side tracking?
Server side tracking can extend the lifetime of cookies by storing information about a 3rd party cookie in a 1st party cookie but Safari’s latest newer versions of ITP have also limited the lifetime of server-side cookies. Using 1st-party data can help brands reduce their reliance on cookies, which would likely have the benefit of being more future-proofed than a solution which looks to circumvent ever-updating cookie regulations and technologies.
What effect have we seen of enforced consent mode?
We’ve seen that Google has limited personalised advertising (remarketing and other audiences) for users who haven’t given their consent and sent a consent signal. We also know that in some cases Google has gone further and disabled conversion tracking and even suspended or terminated accounts completely.
Microsoft Ads has shown similar stricter enforcement in recent months with conversion tracking being disabled for all users without consent signals as well as stopping remarketing lists.
How can businesses balance privacy compliance with the need for valuable attribution data?
When users are expecting/demanding more privacy from their internet experience.
Given the rise in ad blockers and tracking prevention, brands can only access a limited amount of web analytics data, using advanced google consent mode v2 can allow them to model data of users who have opted out of tracking, but in order to get a full picture they may want to competent this with a first-party data strategy and marketing mix modelling.
What’s Next?
Tracking, attribution and general set up is fairly complex and therefore a lot of accounts don’t produce fully trustworthy/accurate data. ie when we start with new clients, we have to clean up their set up.
What do businesses need to prioritise in terms of tracking/attribution?
Brands should work out what they want to be able to track and the limitations that they face based on limitations in their platforms and legal requirements. This can include how measurement works for website events such as ecommerce brands calculating lifetime value, a subscription brand calculating which channel drives the stickiest subscribers or a lead gen site owner calculating where their best leads come from.
This may mean going beyond measurement in just ad platforms and GA4 and combining web analytics with first-party data.
How can businesses overcome the challenges with comparing data historically?
Over the last few years it’s become increasingly difficult to accurately compare data with historical data, changes in user behaviour due to LLMs and AI mode, consent, browser tracking technologies and platform changes have all led to current data being incomparable. To combat this, we need to understand what changes have happened and their impact.
For example, if we know organic clicks are down year-on-year because of AI mode, we might instead look at how our rankings have changed in that time, or what proportion of the clicks we’re getting comparatively to our competitors. If our historical web analytics data isn’t accurate, we may choose to focus on the data we do have and ensure a solid strategy for collecting accurate data going forward.
Have All The Data on Data with Loom
As Lukas has highlighted, while the tools and technologies may evolve, the goal for digital marketers remains the same – to understand audiences, measure effectively and deliver value through data-driven decisions.
Whether it’s adapting to AI-led search, refining attribution models, or building a robust first-party data strategy, the key is agility. At Loom, we continue to help our clients navigate this shifting digital landscape with data and analytics strategies that balance insight, innovation and compliance. For more insight into how we can improve your tracking, get in touch with Lukas today.
