Three signals, one decision

Behaviours, attitudes and expressions all send signals. Knowing which one to believe when they disagree turns data into a decision. 

Every CMO we work with has more data than they know what to do with. When someone finally asks the only question that matters: 'so what do we do?', the ones pulling ahead already know where to look. 

The three jobs every signal does 

Every metric a marketer touches, however new the source, is still doing one of three jobs. Behaviours are the audit trail, the receipts, basically: what people actually did. Attitudes are the compass: what people think and feel and the best hint at where they're headed next. Expressions are the seismograph: every search, post and stray comment, picking up tremors it can't yet tell you are earthquakes. 

Exhibit: the signals map

	What people do, think and say	What technology captures, reflects and generates
Behaviours	Purchases, subscriptions, footfall, repeat purchase, usage, switching	Transaction data, platform conversions, clickstream analytics, e-commerce records, AI-initiated purchases
Attitudes	Brand perceptions, predisposition, trust, perceived value, opinions about quality, category attitudes, recommendation intent, tracked through metrics like Meaningful Difference and Future Power	Nothing analogous to human attitudes, but LLMs and agents already carry brand associations in their training data that shape what they surface and recommend
Expressions	Search queries, social posts, reviews, word of mouth, likes, comments, shares, sentiment	AI recommendations, algorithmic rankings, chatbot answers, generative and synthetic outputs: the emerging 'machine web'

 

Attitudes predict behaviour and we can prove it 

Changes in how people think and feel about a brand reliably arrive before changes in what they do. Kantar’s Future Power measure, built entirely from attitudinal data with no sales input at all, shows why these matters: brands with high Future Power are four times more likely to grow value share than brands with low Future Power. Among the low scorers, 52% were already losing their share. 

Kantar's LINK+ ad-testing system offers marketers a similar head start. On average, ad performance on LINK+ can anticipate shifts in brand equity roughly 11 months before they surface in standard tracking, real advance warning that gives marketers time to act. That timing varies considerably from brand to brand, some effects surface within weeks, others take much longer, but the direction is consistent: strong creative moves the needle on brand equity well before it moves the needle on sales.

When the signals agree, growth compounds 

Octopus Energy is the clearest recent example. In a category known for low satisfaction and high inertia, it built Meaningful Difference through service and green innovation, earned the strongest word of mouth in the sector and rode that to become the UK’s number one electricity supplier in six years. Every signal reinforced the next. 

The pattern isn't one-off. A recent Kantar meta-analysis across six markets found that brand equity can drive up to 35% of new sales. Consistent investment grows further; cutting it lets equity erode and sales stagnate for years. Adidas learned this the hard way: 77% of its budget went to performance marketing, on the strength of dashboards that looked efficient, until the company ran the econometrics and found brand activity was driving 65% of all its sales. 

AI is changing how people discover brands and the stakes are rising. Kantar’s BrandZ data shows perceived brand difference has declined globally for over a decade, from around 18% of brands rated meaningfully different in 2014 to roughly 14-15% today and the proliferation of AI-generated content hasn’t slowed this. More signals were never the goal. Judgement is what sets a brand apart. 

Reading the signals in real time 

Expressions don't just detect a shift is happening, they can diagnose it too. When a leading food brand saw trust scores dip in its tracker, social listening revealed why: a spike in complaints about a recent ingredient change. The survey caught that something was wrong, but it was the expressions caught ‘what’.  

The clearest test of this is a product launch. Two in three new launches are dead or dying in their second year. Most weren’t bad products. The signals were read in isolation or not read at all. 

Exhibit: which signal leads at each stage of a launch

Stage	Signal(s)	What it tells you
Before there's a product	Expressions	Where latent demand is forming
Pre-launch	Expressions	Whether the market is ready
Launch week	Behaviours + Expressions	Whether people are buying, and what they're saying
Post-launch	Behaviours + Expressions + Attitudes	Whether you're building the brand, or just borrowing from it

Before a product exists, only expressions are available: search patterns, cultural tension, early conversation. At launch, purchase data alone can mislead in either direction. It’s only a quarter in, once attitudes are tracked, that you learn whether you built something that lasts or just spiked and faded. 

Stop measuring everything. Start reading what matters.

This means knowing which signal to trust and having the nerve to act before all three agree. That’s the case we make in full in Three Signals, our latest paper on moving from signal overload to the confidence to act. It sets out the complete signal hierarchy, the patterns that show up when signals diverge and a practical guide to matching the depth of evidence to the size of the decision in front of you. 

Download the full Three Signals paper to build your own signal hierarchy or speak to someone at Kantar about turning your data into signal and your signal into action.