Advertising Effectiveness Research: What the Data Tells You

Advertising effectiveness research is the practice of measuring whether your advertising is working, why it is working, and what you should do differently. Done well, it connects spend to outcomes across the full purchase funnel, separates correlation from causation, and gives decision-makers a defensible basis for budget allocation. Done poorly, it tells you what already happened and calls it a strategy.

Most marketing teams sit somewhere between those two states. They have data. They have dashboards. What they often lack is a coherent framework for interpreting it honestly.

Why Most Effectiveness Research Gets Misread Before It Is Even Analysed

Early in my career I was deeply invested in lower-funnel performance metrics. Click-through rates, conversion rates, cost per acquisition. The numbers were clean, the feedback loop was fast, and the narrative wrote itself. Performance was working. Brand was fuzzy and hard to measure. So we leaned into performance.

It took years before I started asking a harder question: how much of what performance marketing was “driving” was going to happen anyway? Someone who searches for your brand name after seeing your TV ad is not a performance conversion. They are a brand conversion that got counted in the wrong column. The performance channel got the credit. The brand investment got the budget cut.

This is not a niche problem. It is one of the most common distortions in advertising effectiveness research, and it shapes budget decisions at every level of the market. The clothes shop analogy is useful here: someone who has already tried something on is far more likely to buy than someone browsing cold. Performance marketing often intercepts people who have already been warmed by brand exposure. The intercept gets measured. The warming does not.

If you are building a go-to-market strategy that is meant to drive sustainable growth, understanding where advertising effectiveness research fits, and where it breaks down, is not optional. The broader principles behind that kind of commercially grounded planning are covered in depth across the Go-To-Market and Growth Strategy hub.

What Advertising Effectiveness Research Is Actually Measuring

Before you can interpret effectiveness data, you need to be clear about what the research is actually designed to measure. There are four distinct questions that advertising effectiveness research can answer, and they require different methodologies.

The first is exposure: did people see the advertising? Reach, frequency, and impression data answer this. It is a necessary starting point but tells you nothing about whether the advertising did anything useful.

The second is recall and recognition: do people remember seeing the advertising, and can they connect it to the brand? Brand tracking surveys, ad recall studies, and awareness metrics sit here. This is where a lot of traditional brand measurement lives.

The third is attitude and perception: has the advertising shifted how people feel about the brand? This requires pre and post measurement, usually through survey-based brand health tracking. It is harder to do well and slower to move, but it is often the most predictive of long-term commercial outcomes.

The fourth is behaviour: did the advertising cause people to do something different? Purchase, trial, search, visit. This is where performance data, sales uplift analysis, and incrementality testing live. It is also where most of the misattribution happens.

The mistake most teams make is treating one of these as a proxy for all four. Measuring recall and assuming it means the advertising is driving sales. Measuring conversions and assuming it means the advertising is creating demand. Each layer of the funnel requires its own measurement approach, and none of them should be read in isolation.

The Attribution Problem Nobody Has Solved

I have spent a significant part of my career working in and around attribution. At iProspect we managed substantial media budgets across dozens of clients, and the attribution conversation came up constantly. Every client wanted to know which channel was working. Every channel partner claimed credit. The models we used were sophisticated. They were also, in important ways, wrong.

Attribution models, whether last-click, first-click, linear, time-decay, or data-driven, all share the same fundamental limitation: they describe the path a conversion took, not the cause of the conversion. A customer who clicked a paid search ad after seeing three display impressions, a YouTube pre-roll, and a direct mail piece did not convert because of the paid search click. They converted because of a sequence of exposures that the model is not equipped to weight correctly.

Data-driven attribution has improved on the older heuristic models, but it still operates within the walled gardens of individual platforms. Google’s data-driven attribution model is trained on Google data. It has no visibility into what happened on Meta, in a store, on a podcast, or on a billboard. The model is not lying to you. It is just answering a narrower question than the one you are asking.

The practical implication is that attribution data should inform channel investment decisions at the margin, not determine them wholesale. Use it to understand relative performance within a channel mix. Do not use it to conclude that one channel is doing all the work while others are redundant. That conclusion is almost always wrong, and acting on it tends to erode the brand foundations that make performance channels efficient in the first place.

Incrementality Testing: The Most Honest Tool You Are Probably Underusing

If attribution is the question of which channel got credit, incrementality is the question of what would have happened without the advertising. These are different questions, and incrementality is the more useful one.

An incrementality test, at its simplest, involves exposing one group to advertising and withholding it from a control group, then measuring the difference in outcomes. The lift you observe is the incremental effect of the advertising. What would have happened in the control group represents baseline behaviour that the advertising cannot claim credit for.

This matters because a significant portion of the conversions that performance channels report as wins are not incremental. They are people who would have purchased anyway, who happened to click an ad on their way to completing a transaction they had already decided to make. The ad got counted. The purchase was not caused by it.

Incrementality testing is not new, and it is not technically out of reach for most teams. Geo-based holdout tests, platform-native lift studies, and matched market testing are all accessible methodologies. What holds most teams back is not capability but willingness. Running an incrementality test means accepting the possibility that a channel you have been investing in heavily is less effective than you thought. That is a politically uncomfortable result to present.

I have been in those rooms. The test results come back showing that 40% of reported conversions were not incremental, and the first instinct is to question the methodology rather than revise the budget. Understanding the difference between what measurement tools can prove and what they cannot is central to how effective teams make decisions. It is also one of the reasons why go-to-market execution feels harder than it used to: the measurement infrastructure has not kept pace with the complexity of modern media environments.

Brand Tracking: What It Can and Cannot Tell You

Brand tracking is one of the oldest tools in advertising effectiveness research, and one of the most frequently misunderstood. A well-designed brand tracker measures awareness, consideration, preference, and perception over time, giving you a longitudinal view of how advertising is shifting the brand’s position in the market.

The problem is that most brand trackers are designed to confirm that advertising is working, not to test whether it is. They measure metrics that are sensitive to advertising exposure, such as prompted awareness and ad recall, and report movement on those metrics as evidence of effectiveness. But prompted awareness of an ad is not the same as a shift in brand preference, and brand preference is not the same as commercial outcomes.

A more useful brand tracking programme measures metrics that are genuinely predictive of future purchasing behaviour. Spontaneous brand awareness, consideration set inclusion, and relative preference against named competitors are more commercially meaningful than ad recall scores. They are also slower to move, which is why impatient organisations tend to default to the faster-moving metrics that tell a more immediately satisfying story.

Brand tracking data becomes genuinely powerful when it is read alongside sales data over a long enough time horizon. Short-term sales data and brand health metrics often move in opposite directions. A brand that cuts advertising spend will typically see short-term sales hold up while brand metrics quietly deteriorate. By the time the sales impact becomes visible, the damage has been compounding for months. This is the core argument for maintaining brand investment through downturns, and it is an argument that brand tracking data can make, provided the data goes back far enough and is read honestly.

Marketing Mix Modelling: Useful, Expensive, and Often Misapplied

Marketing mix modelling, sometimes called econometric modelling, uses regression analysis to decompose sales into their contributing factors: base sales, promotional activity, media spend, pricing, distribution, and external variables like seasonality and economic conditions. Done well, it is the most comprehensive view of advertising effectiveness available.

The limitations are real and worth naming. MMM requires substantial historical data, typically two or more years of weekly data across all relevant variables. It is expensive to build and maintain. The model outputs are only as good as the inputs, and if your media spend data is incomplete or your sales data is noisy, the model will produce confident-looking numbers that are based on shaky foundations.

I have seen MMM outputs used well and used badly. Used well, they inform long-range budget allocation decisions, help identify diminishing returns thresholds for individual channels, and provide a basis for scenario planning. Used badly, they become a post-hoc justification for decisions that have already been made, or a way of burying inconvenient findings in statistical complexity that most stakeholders will not interrogate.

The most important thing to understand about MMM is that it models the past. It tells you how your advertising has performed under the conditions that existed during the modelling period. It cannot account for changes in competitive intensity, media landscape shifts, or changes in consumer behaviour that fall outside the historical data. Treat it as one input into a planning process, not as a decision engine.

For teams that cannot justify the investment in full econometric modelling, lighter-touch approaches like media mix analysis using platform data combined with geo holdout tests can provide a reasonable approximation. The goal, as with all effectiveness measurement, is honest approximation rather than false precision.

Qualitative Research: The Part of Effectiveness Most Teams Skip

Quantitative effectiveness research tells you what is happening. Qualitative research tells you why. Most teams invest heavily in the former and neglect the latter, which means they end up with a lot of data about outcomes and very little understanding of the mechanisms driving them.

I remember being in a brainstorm for Guinness early in my career. The brief was built on solid quantitative data about brand perceptions and purchase behaviour. What it was missing was any real understanding of why people felt the way they did about the brand. The numbers described the situation. The qualitative work, when we did it, explained it. Those are different things, and the creative work that comes from understanding the why is almost always stronger than the work that comes from optimising against the what.

Qualitative research in an effectiveness context includes focus groups and depth interviews to understand how advertising is being received, ethnographic observation to understand how purchase decisions are actually made, and social listening to identify organic consumer language and sentiment. None of these replace quantitative measurement, but they provide the interpretive context that makes quantitative data meaningful.

Tools like Hotjar can provide behavioural data on how people interact with digital touchpoints, which sits at the intersection of quantitative and qualitative insight. Session recordings and heatmaps do not tell you why someone left a page, but they can surface friction points that prompt the right qualitative questions.

Building an Effectiveness Measurement Framework That Works

The practical challenge for most marketing teams is not a shortage of measurement options. It is the absence of a coherent framework that connects the different measurement approaches to each other and to business outcomes.

An effective measurement framework starts with a clear view of the business objective. Not the marketing objective. The business objective. Revenue growth, market share gain, customer acquisition at a target cost, retention improvement. The marketing activity and its measurement should be traceable back to that objective, not to a proxy metric that marketing teams find easier to move.

From the business objective, you work backwards to identify the behavioural outcomes that advertising needs to drive, the attitudinal changes that precede those behaviours, and the awareness and exposure conditions that need to be in place for attitude change to occur. Each layer requires its own measurement approach, and the approaches need to be read together, not in isolation.

The framework also needs a time dimension. Short-term effectiveness and long-term effectiveness are not the same thing, and optimising for one at the expense of the other is one of the most common and costly mistakes in advertising investment. A campaign that drives strong short-term sales response while eroding brand distinctiveness is not performing well. It is consuming brand equity that took years to build.

Commercial transformation at scale requires this kind of joined-up thinking. BCG’s research on commercial transformation identifies measurement capability as one of the core enablers of marketing-led growth, alongside talent and technology. The measurement piece is often the last to receive serious investment, which is why so many organisations have sophisticated media capabilities sitting on top of a measurement infrastructure that was designed for a simpler media environment.

For teams that want to sharpen their go-to-market thinking more broadly, the principles covered here connect directly to the commercial planning frameworks explored across the Go-To-Market and Growth Strategy hub.

What Effie Judging Taught Me About Effectiveness

Judging the Effie Awards gives you a particular perspective on effectiveness research that is hard to get anywhere else. You read hundreds of case studies from brands that have done the work to prove their advertising drove commercial outcomes. You also develop a strong instinct for when the proof is genuine and when it is constructed after the fact to fit a narrative.

The cases that stand out are not the ones with the most impressive numbers. They are the ones where you can see a clear line of reasoning from objective to strategy to execution to outcome, supported by measurement that was designed before the campaign ran rather than assembled afterwards. Pre-defined KPIs, baseline measurements taken before launch, control groups where possible, and a willingness to report what did not work alongside what did.

The cases that do not hold up tend to share a common characteristic: the measurement was designed to prove a conclusion rather than test a hypothesis. The metrics were selected because they moved in the right direction, not because they were the right metrics. The competitive context was ignored. The base rate was not established. The result looked impressive until you asked the question that the case study was carefully avoiding: would this have happened anyway?

That question, “would this have happened anyway,” is the most important question in advertising effectiveness research. It is also the one that most measurement frameworks are not designed to answer. Building your measurement approach around that question, rather than around the question of how much activity you generated, is what separates effective measurement from measurement theatre.

Understanding how revenue teams are thinking about pipeline and measurement is also shifting. Vidyard’s Future Revenue Report highlights how go-to-market teams are increasingly aware of the gap between reported pipeline and genuinely incremental opportunity, a tension that mirrors the incrementality problem in advertising measurement almost exactly.

About the Author

Keith Lacy is a marketing strategist and former agency CEO with 20+ years of experience across agency leadership, performance marketing, and commercial strategy. He writes The Marketing Juice to cut through the noise and share what works.

Frequently Asked Questions