More marketing teams are now thinking about incrementality than ever before – which we’re obviously very excited about. They’re realizing that they need to move beyond multi-touch attribution and that there are very meaningful questions that platform-reported dashboards and last-click attribution just aren’t built to answer. Again, this is great news for the marketing ecosystem.
But incrementality is not an easy journey for most companies. The measurement status quo is very ingrained in their culture, mostly because they provide familiar, legible answers that are easy to explain in an exec meeting. And we get it – many finance teams and senior executives just want clean numbers and clear accountability, and it can be very complicated to introduce other ways of measurement that deal with uncertainty and causality.
Changing a company’s culture is extremely difficult. We’ve seen it often when it comes to media mix modeling – maybe there’s a champion that wants to move their org from MTA to MMM, but their team still hasn’t fully internalized the blind spots of their current system.
That’s why many MMM initiatives stall before they start. The organization hasn’t yet internalized the idea that correlation is not causation. And until that clicks, no amount of modeling sophistication will change how decisions actually get made.
Before you introduce advanced measurement, you need to change how people think about cause and effect in marketing. And we recommend starting with brand search.
Branded Search: The Easiest Way to Prove a Hard Concept
Branded search is one of the most consistently over-credited channels in marketing – and that’s exactly why it’s so useful.
In a last-click attribution report, branded search almost always looks amazing. The customer types in your brand name, clicks your ad, and converts. The platform reports a 10x ROAS, and the team celebrates.
But intuitively, most people understand what’s actually happening: the customer was already going to buy. They just clicked the sponsored link instead of the organic one.
When you run a brand search test – let’s say you pause for 2 weeks – two things can happen:
- Conversion volume stays flat: This tends to happen when your brand is already well known, you’re the clear leader in your space, our products are unique and not easily substituted, and competitors aren’t aggressively bidding on your brand terms.
With this, now you can go to your finance and exec team and show them with hard data that that brand search is not as incremental as the platform reported. This is often the “aha” moment that shows them the difference between attribution and actual business impact. For teams new to incrementality, it’s often the first time they see the difference between what got credit and what actually drove the outcome.
A classic example is eBay. Years ago, they paused branded search and saw no drop in traffic. People weren’t discovering eBay through ads – they were using Google as a navigational tool. The conversions would’ve happened anyway.
- Conversion volume drops heavily: Brand search might actually be incremental – especially if you operate in a crowded category, your product isn’t differentiated, and competitors are bidding on your brand terms. In these cases, branded search might be doing real defensive work. If someone searches your name and sees a competitor offering 15% off in the top slot, they might steal the sale.
But even if brand search is as incremental as you thought, that’s also good news. You can still go to your exec team and show them the data, propose a budget increase on the channel because of it, and explain what it would’ve meant if the results had been the opposite.
Whether or not branded search is incremental is not just a yes or no question, and it’s going to vary from company to company and vary over time. What matters is that you test it and start building an experimentation culture around incrementality.
And because it’s a well-known channel with relatively low perceived risk, it’s a much easier political lift than running a full-blown geo test or pausing TV.
How to Design a Branded Search Test That Actually Works
Another reason why brand search is such a good place to start is that it’s relatively easy to run. You don’t need a year-long study or advanced modeling todo it well. All you need is a clean setup and clear metrics.
You have three main options:
- Full pause – Temporarily turn off branded search ads. This gives the cleanest signal, but it’s also the riskiest politically.
- Spend up and down – Dial branded search spend down or up significantly. Still provides a strong signal with less perceived risk.
- Geo-split – Turn off branded search in a subset of markets and leave it on in others. This is often the best option for large brands that can’t pause nationally.
Each approach should be paired with a tight testing window (e.g., 1–2 weeks), clearly defined KPIs (e.g., conversions, revenue, new customer acquisition), and a stable baseline. Avoid running during major promos or holidays because seasonal volatility will muddy the results.
What you’re measuring is simple: what happens to sales when you stop paying for branded clicks? The difference between your control and treatment groups is the lift or lack thereof.
And once you’ve got the results, walk the team through it. Show them that high last-click ROAS doesn’t always mean incremental revenue.
The real value of a branded search test isn’t the result itself. It’s what happens after people see it. You can use this example as an intuitive gateway to help them think about running experiments to validate channels, and how MMM might fit into the picture to understand what your marketing is actually doing.
TLDR:
- Most orgs struggle to adopt incrementality thinking because they’re anchored to attribution. Before introducing MMM, you need to change how people think about cause and effect.
- Branded search is the perfect first test. It’s consistently over-credited in last-click attribution, easy to run, and low-risk politically – making it an ideal teaching moment.
- Run a simple test (pause, dial down, or geo-split for 1-2 weeks) and measure what happens to conversions. Either brand search isn’t as incremental as platforms reported, or it is – both outcomes teach the lesson.
- The real value isn’t the result itself. It’s building an experimentation culture and showing your team the difference between what got credit and what actually drove the outcome.
