Google Marketing Live 2026 covered a lot of ground – search, YouTube, commerce, agentic AI. But running through almost everything Google announced was the same underlying theme. The quality of your data has become one of the most significant factors in how well your advertising performs.
This isn’t a sudden shift. Data management has been growing in importance for years. What GML 2026 made clearer is that the trend isn’t slowing down – and for businesses that haven’t started taking it seriously, the gap between them and those that have is widening.
The algorithm needs teaching
It helps to understand how Google Ads actually works before talking about data. The platform runs on machine learning. It makes decisions about who to show your ads to, when, and at what bid, based on the signals it’s trained on. Those signals come from your account – your conversions, your audience data, your customer lists, your feed quality.
Think of it like training a puppy. When the puppy does something right, you give it a treat. Over time, it learns to repeat those behaviours. Each piece of data you feed into Google Ads works in a similar way – it’s a signal to the algorithm about what a good outcome looks like for your business. Feed it the right signals consistently, and it learns to find more of the customers you actually want. Feed it inconsistent or poor-quality data, and you end up with something that’s difficult to manage and expensive to run. The difference between those two outcomes is largely – though not entirely – about data quality.
What Google said at GML 2026
The commercial case is reasonably well evidenced. Advertisers using first-party data – data collected directly from their own customers and website visitors, rather than sourced from third parties – see an average 11% incremental return on ad spend increase, according to Google. Businesses that have properly implemented Google Tag Gateway, a tool that improves how conversion data flows into the platform, see an average 14% increase in measured conversions.
In our experience working across client accounts, the improvement can be meaningfully higher than those averages for businesses where data quality has been genuinely poor. Getting the foundations right doesn’t always produce dramatic overnight results, but it tends to compound over time as the algorithm has more to work with.
The 27% conversion increase Google attributes to AI Max – its AI-powered campaign expansion tool – is worth noting in the same breath. That figure depends on the account being properly configured to receive those results. If your tracking is incomplete or your signal quality is low, the tool will still spend your budget. It just won’t do it as effectively as it could.
Three issues we see most often
We look at a lot of accounts, and the data problems tend to fall into three categories.
- The first is incorrectly configured tags. Conversion tracking that fires on the wrong pages, double-counts events, or misses key actions. The downstream effect on campaign optimisation can be significant, and it’s often invisible until someone actually audits the setup.
- The second is a limited understanding of consent mode and cookie policies. Since GDPR and the shift in how browsers handle third-party cookies, the rules around data collection have changed considerably. Consent mode needs to be properly implemented to allow Google’s modelling to work in the gaps where users don’t grant consent. Without it, you’re losing signal – and the loss is easy to miss because nothing obviously breaks.
- The third is attribution gaps. If you can’t reliably connect an ad click to a meaningful outcome, the algorithm ends up optimising for whatever it can measure rather than what actually matters to your business.
The CRM opportunity
One area where a lot of SMEs are still leaving value on the table is the CRM – customer relationship management software, the system that stores and manages your customer and lead data.
The limitation of relying solely on event tagging is that it treats all conversions equally. A form submission is a form submission, regardless of whether that lead became a paying customer or went nowhere. If your account is optimising for conversion volume without any quality filter, it’ll find more of everything – including leads that don’t convert.
A CRM with lead scoring allows you to take that data offline, assess which leads actually turned into revenue, and feed that qualified information back into your ad platforms. The algorithm can then learn what a genuinely good lead looks like for your specific business. That’s a more useful signal than raw conversion volume, and the difference in account performance over time can be considerable.
This kind of offline data integration matters more now than it did a few years ago. Most businesses used to rely on event tracking because that was largely what the platforms supported and encouraged. The tools have moved on, and so have the standards for what well-managed data looks like.
AI needs context, not just data
One of the more interesting announcements at GML 2026 was Google AI Brief – a feature that lets advertisers provide natural language descriptions of their brand voice, guidelines, and guardrails, which then shape how Google’s AI generates creative content.
It’s a useful illustration of a broader point. Google’s advertising AI is trained on aggregate data across billions of interactions, which makes it capable and generally effective. But its default output reflects the average across everyone using the platform. That average is a reasonable starting point. It’s not a substitute for training the system on your specific business.
First-party customer data, lead quality signals, audience lists, product feed data, brand guidelines fed into AI Brief – all of it serves the same purpose. The more context you give the system about what good looks like for your business, the less it has to rely on what good looks like in general. This is the direction Google is clearly moving in, and most of the new features announced at GML 2026 sit within that same logic.
Where it makes sense to start
Before looking at account structure, ad copy, or creative, it’s worth taking stock of your data. A few questions worth asking:
Tagging and measurement. Are your conversion events configured correctly? Do you have consent mode set up properly? Are there gaps in your attribution you haven’t accounted for?
Lead quality. If you’re running lead generation campaigns, are you feeding quality signals back into the platform, or is the algorithm optimising on raw volume?
First-party data. Do you have customer lists, CRM data, or purchase history that could improve targeting? If it’s sitting in a system that isn’t connected to your ad accounts, it’s not contributing anything.
These aren’t minor technical details. They’re foundational – and addressing them is often where the most straightforward performance improvements come from. The signals from Google Marketing Live 2026 suggest this is only going to matter more over time, not less.
The Last Word
Data quality has always mattered in Google Ads. What’s changed is how much the platform now depends on it, and how clearly Google is signalling that this dependency will only deepen. Whether it’s fixing gaps in your tracking, connecting your CRM to your ad accounts, or giving AI tools the context they need to work properly for your business rather than the average of everyone else’s, the foundations are what make the difference. If data management is something you’d like to get right, get in touch with our team at Herdl – it’s one of the first things we look at, and often where we find the most room to improve.