Is It Time We Talked About Turning Off Your Ads? (And Yes, You Read That Right)

Your best-ever ROAS month can still lose you money. Brenden Delarua explains why the number on your ad dashboard rarely matches the number in your bank account, and what to measure instead. Summary Brenden Delarua spent 11 years in paid media, hitting and smashing ROAS targets across Meta, Google and Criteo, until a CFO asked him a simple question at a quarterly review. If the ad numbers are this good, why is revenue down? That question sent him into causal measurement, and he now runs Stella (Growth Intelligence), a marketing-measurement company that helps brands work out what is actually driving sales rather than what the platforms claim. In this conversation Brenden unpacks why platform-reported ROAS is a correlation metric that can overstate how well your ads are working, and how chasing it pushes marketers into retargeting and branded-search quick wins that flatter the dashboard without growing the business. He walks through the practical ladder any small brand can climb, from multi-touch attribution and post-purchase surveys up to full geo-holdouts and media mix models. Along the way there is a YouTube campaign that looked like a disaster, a lesson on why branded search is not always what it seems, and the one post-purchase survey question worth more than the rest. ROAS Is Not the Same Thing as Profit Brenden's whole approach traces back to one uncomfortable review meeting. As a media buyer he had beaten every ROAS goal he was set, then presented record-breaking results, and still got questioned by the CFO. "If you're showing us that ROAS is so high, why am I not seeing that in our actual bank account?" — Brenden Delarua, on the CFO question that started it all The problem is what ROAS actually measures. It is a correlation metric. Someone saw or clicked an ad and later converted, so the platform claims the sale. "ROAS is not correlated with profit, even though a lot of people think it is." — Brenden Delarua Because ROAS rewards clicks, it tends to deprioritise channels you cannot click, like connected TV or audio, and it tempts marketers to pile budget into retargeting or branded search for a quick win. The dashboard looks great. The business stops growing. Brenden's point is that what owners ultimately care about is profit, not a platform-reported number. The YouTube Campaign That Looked Like a Money Pit One story makes the gap between platform numbers and reality concrete. Brenden runs ads for a client spending heavily on YouTube, a channel infamous for a poor click-to-convert ratio. In-platform, the YouTube ROAS sat at around 0.2, so the account ROAS looked like it was suffering. The obvious call was to question why so much was going into a channel that appeared to be losing money. Then he pulled the post-purchase survey data from NoCommerce, and around 7% of buyers said they came from YouTube. Set against the spend, that share of revenue worked out at roughly a 3x return, and the channel was actually very profitable. The lesson is not that YouTube always works. It is that a single click-based number can hide the truth, and a second data point, in this case a qualitative survey, can change the whole decision. "Sometimes increasing incremental ROAS is the opposite of increasing platform ROAS." — Brenden Delarua What is Causal Measurement? Brenden is keen to stress that incrementality is not just a big-brand concern, and you do not need to turn off ads to benefit from the thinking. He lays out a practical ladder that scales with the business. Start with proper multi-touch attribution. For e-com tools he rates Triple Whale and Northbeam. It maps the customer journey by stitching anonymised touchpoints together until someone converts, then shows first touch, last touch or a fairer linear model. Add post-purchase surveys. Even a brand-new store can ask buyers where they came from and, more importantly, why they bought. Tools like NoCommerce can use conditional logic to ask which specific ad someone remembers. Add causal analysis when you are big enough. Once sales volume is there, usually for brands over roughly 10 million a year running on two or three platforms, holdout testing and media mix models come into play. "You don't have to do holdouts to understand that what's causing sales to happen might be different from what you're seeing in ad platforms." — Brenden Delarua For a brand doing a million a year, the same issues exist at a smaller scale. The trouble is that as you grow, working out what is really driving growth gets harder, not easier. What are iROAS and Holdouts? For brands ready to test properly, Brenden explains the terms plainly. A holdout study means turning ads off in certain geographical regions and watching your source-of-truth revenue, usually Shopify, against the regions that stayed live. The measure that comes out of it is iROAS. "iROAS stands for incremental return on ad spend, every dollar we put in that gives us back money that would not have happened without that investment." — Brenden Delarua A few practical points from the conversation: Test in a cycle, from channel level (all of Meta, all of Google) down to tactic level (branded versus non-branded) and then individual campaigns, then start again. Holdouts typically run 20 to 30 days, with top-of-funnel channels like CTV closer to 45. Branded search is often called non-incremental, but not always. If the brand name is effectively the product category, it can be highly incremental. Matt's own analogy landed it well, as if a company selling omega-3 were simply called Omega-3. Media mix models look back over around two years, factor in seasonality and other variables, and are forward-looking, so they can forecast how to allocate next quarter's budget. Brenden recommends using holdouts and media mix models together, with the holdout calibrating the model. There is a catch, and Brenden names it. A holdout is a snapshot in time, shaped by your bid strategy, the platform algorithms and how many competitors are in the market that month. His ads in February behave nothing like his ads in the run-up to Black Friday. "The moment you get that iROAS report back, the next day it starts to shift. We call it causal decay." — Brenden Delarua The Post-Purchase Question You Should Ask Asked for his best parting tip, Brenden returned to post-purchase surveys and one question in particular, credited to Jarrell Blades, VP of Growth at Tushy. "I love asking what almost stopped you from converting, because that gives you tons of insight into your messaging and targeting." — Brenden Delarua His advice is to begin with the end in mind, borrowing the idea from The 7 Habits of Highly Effective People, and design the survey around the answers that will actually help you convert more people. Surveys surface the things attribution never sees, like a friend's recommendation or word of mouth. They also reveal which ad someone remembers, which is not always the ad they clicked. When the most memorable ad differs from the clicked one, that tells you something powerful about which creative is really doing the work. Today's Guest Today's guest: Brenden Delarua Company: Stella (Growth Intelligence) Website: https://www.stellaheystella.com/ LinkedIn: Connect with Brenden on LinkedIn TikTok: @brendanbuilds (the lighter, more sarcastic side of his marketing content) Resources Mentioned Triple Whale and Northbeam — multi-touch attribution tools Brenden recommends for e-com NoCommerce — post-purchase survey tool, including conditional logic on higher plans Jarrell Blades, VP of Growth at Tushy — source of the "what almost stopped you from converting" survey question Neil Hoyne, former Google data specialist and author of Converted — a previous EP guest, referenced for the shopper whose single purchase involved 236 touchpoints The 7 Habits of Highly Effective People — for the "begin with the end in mind" principle applied to survey design Keep Listening Find the full archive, free tools and everything else at ecommerce-podcast.com. New episodes land every week. If you are getting value from the show, subscribe on your podcast player of choice and share this episode with a fellow founder who is still judging their ads by the dashboard alone. Episode link: https://www.ecommerce-podcast.com/is-it-time-we-talked-about-turning-off-your-ads-with-brenden-delarua