Newton Baby is a baby mattress and product manufacturer that brings an appetite for incrementality and a test-heavy diet to to its marketing.
Aaron Zagha, Newton Baby’s CMO, has a different approach than other DTC companies, let alone mattress makers.
“The whole mattress business was a whole interesting clusterf*** for a while, of which many college textbooks have written,” Zagha tells AdExchanger.
One significant differentiator between Newton Baby and its competitors, is that Newton never took outside money, he says. “My mandate for a while was to grow as fast as possible while holding breakeven.”
That was until a couple of years ago, when the business turned to profitability.
And profitability is now being increased, Zagha says, by wringing greater efficiency from marketing, with help from marketing science platform Haus. In the past year, Newton Baby has put all its media channels through a new testing regime for incrementality – i.e., did this ad budget bring new shoppers to the brand who otherwise wouldn’t have made the purchase?
It was a revelatory experience.
AdExchanger: What is your paid media mix?
AARON ZAGHA: Our split right now is basically 50-40-10, in terms of our own site, Amazon and Babylist [a large registry and online baby product retailer]. We do a little brick and mortar in Canada. Next year, that’s going to change. We’re into a much bigger retail rollout.
We’re essentially DTC, though, so about 100% of our marketing dollars are online.
At any given point, we run 20 to 25 channels – half of which are permanent, half are test and learn. And in order of spend, it’s influencer, Google, Facebook, Amazon, TikTok, Pinterest, and down from there.
What’s “influencer” doing up there?
Good question. We’re for sure the biggest influencer spender in the baby space.
Our core demographic is a first-time parent who’s looking for advice. Across the board, it’s terrifying having a kid. It certainly was for me. And if people you like and respect on social media recommend a product that they used for their baby, I think you’re much more likely to emulate that purchase.
I hadn’t had much success with influencer, actually, at prior roles. But it made all the sense in the world for Newton.
And, like everything else, we have a test-and-learn approach. Influencer worked for us. So we went in pretty heavily.
What vendors do you use for that test-and-learn measurement?
We’ve been working with Haus for a year now on incrementality testing and with Rockerbox for five years now on multitouch attribution and media mix modeling.
Why both?
It’s been a journey for us, as we’ve gotten to scale.
You learn about the nuances and, most importantly, see the vast differences in results.
It’s hard to argue against incrementality and what Haus does, because it’s causal. Everything that isn’t incrementality testing is correlation – multitouch attribution and media mix modeling are both correlations. The problem with correlation is that if everyone who makes a purchase ends up seeing a branded search ad, branded search looks super important in driving business to your site.
But we now know, in our case, it’s zero.
Zero? Or did sales dip while converting more profitably?
You might think that. But it’s zero.
The reason I hired Haus was because I met Zach [founder and CEO Zach Epstein] at a conference, and we were talking shop. He asked how much I spend on branded search. And I told him.
He said, “I’ll bet you that you should be spending zero.”
And I was like, “Get out of here.”
So we bet on it, with a contract out if he proved right. And, you know, he was right. The savings on branded search on Google paid a 20X return on Haus’ fees.
Everyone who searches for Newton Baby is going to buy our product anyway, regardless of whether or not they click the first ad they see or scroll down a little bit to find the link.
It’s definitely true of Google. Amazon we’re less sure.
Is Amazon branded search more effective?
Probably not. It’s very hard to do incrementality testing for Amazon sponsored search.
How so?
You can’t do geo-holdout groups in Amazon.
It’s not Amazon search results so much as you can’t effectively run an experiment?
Right.
You have to design an experiment that’s going to result in statistically significant results. And with holdout groups that you can ensure didn’t run campaigns, to see what happens at different spend levels.
That can’t be done with causal measurement on Amazon right now.
Will your media mix change after you’ve done a brick-and-mortar rollout next year?
I don’t expect it to. I think the way people discover brands now isn’t in shopping aisles; it’s social media.
And we’re in a very different point in our growth cycle, versus when older DTC startups would go into stores for the first time, doing giant end-caps and paying all kinds of fees for promotion. We’re not doing any of that.
And, again, with Haus, we’ll geo-lift test the stores where we launch.
Do you use – and trust – the incrementality measurement solutions from Google and Meta?
Facebook has an incrementality testing product, and Google has one for YouTube, but not Search.
We use them, but the problem is they can’t take into account the rest of your spend and what’s going on in your business. They only see their channel.
That being said, results we’ve seen from Haus tests are similar to what we saw from running limited Facebook incrementality tests.
But, also, do you trust the platforms on the reporting? There’s always that question, of their incentives being very different from ours.
With so much testing – multiple vendors, half your media channels for testing, the cost of testing itself – are you tempted to fly blind, so to speak, and spend more on media?
Every dollar I spend has an ROI hurdle, including paying for tools and for potential losses of revenue on a failed test.
If we’re doing a geo-holdout test, we’re losing revenue. We’re not showing ads in some part of the country. We just have to make sure the returns exceed our hurdles.
But given the continued rise in CPMs, we have to continue testing.
And not only CPMs, but algorithms changing, too. Facebook’s efficacy one month can be wildly different the next month. And unless you’re testing, you won’t know what’s really impacting your business.
And, most importantly, how you should be budgeting.
This interview has been lightly edited and condensed.