Hey!

After spending eight figures on paid ads at D.LOUISE, the most expensive lesson I've learned is that the ads with the worst ROAS in your account are often the ones holding everything else up.

Back in 2022, founder ads took the majority of our ad spend. Over the last couple of years we watched their in-platform ROAS deteriorate, so we did what the numbers seemed to be telling us and started pulling them back, along with our other top-of-funnel ads that explained the brand (usually longer-form videos about what makes the product different). Almost immediately, the whole account started to pull down.

What we'd missed is that you can't just look at which ads are performing in the account and switch off the ones that aren't. Each ad has a job, and the lens you need to look through is a full funnel one: how do you turn someone who has never heard of you into a customer who is convinced enough to buy?

In a category like jewellery, where brand and perception matter so much, that is never going to happen through a single static ad. It takes multiple touchpoints, and it goes beyond the ads themselves, right down to the aesthetic of your imagery and what someone sees when they click through to your Instagram feed.

Why your bottom-of-funnel ads always look like the heroes

Depending on how your attribution is set up, your bottom-of-funnel ads will almost always show the highest ROAS, because they're the last thing someone sees before they buy.

Black Friday is the perfect example. We could put up a static ad saying 60% off and it would take a considerable amount of spend at a great ROAS. But the only reason it performs so well is all the ads that ran before it. On its own, that ad would never convince someone who had never heard of D.LOUISE to buy.

The metric we look at instead

One metric we've started to lean on is new visitor percentage (NVP): the share of people clicking through from an ad who have never been to your site before. It tells you whether an ad is actually bringing new people into your funnel, or just collecting people who were already on their way to buying.

Funnel showing top-of-funnel ads with low ROAS but high new visitor percentage

When you look at an account through this lens, the picture changes. A top-of-funnel ad running at 0.9 ROAS with 74% new visitors is doing a very different job from a bottom-of-funnel ad at 4.1 ROAS with 12% new visitors. Switch off the first one and, a few weeks later, the second one has nobody left to sell to. So leave your high NVP ads on, find the ones with the highest NVP, and make more like them.

How I'd test if I was launching a brand today

If I was starting from scratch, I'd begin by mapping out the different angles I could use to sell the product. Say I was launching a brand that helps with male baldness. One angle might be: "Do you want to feel more confident in front of women?"

For each angle, I'd build a set of top, middle and bottom-of-funnel ads in Figma. At the top, video that explains the angle and the problem. In the middle, proof that it works, like reviews, before and afters or a demonstration. At the bottom, an offer or a guarantee, while still staying laser-focused on that same angle.

One ad angle built out as top, middle and bottom of funnel ads

Then, rather than judging each ad on its own, I'd look at the data for the angle as a whole: its combined performance, and how much new traffic it's bringing in. These ads don't work in isolation, so they shouldn't be judged in isolation either.

Get out of the ad account

Your ad account is only one data point. At D.LOUISE we run our post-purchase surveys through Grapevine, and it's one of the simplest ways we've found to understand the real journey from stranger to customer. A few well-chosen questions will tell you things no attribution tool can, including how long your consideration window really is.

Three post-purchase survey questions that map the customer journey

If you ran a shop, you'd be standing next to your customers every day and you'd learn all of this naturally. In e-commerce we're hidden from these core things, so you have to go out of your way to find them.

Jack

P.S. Next time you're about to switch off an ad because of its ROAS, check its NVP first. It might be the ad paying for all the others. And if you'd like me to look at your own funnel with you 1:1, you can apply here.