Usually, we talk about science in terms of biology, tech, medicine, or climate. But we shouldn’t overlook the scientific approach necessary to succeed in DTC these days. 

It takes volume and rigorous methods to shed light on what works in the digital halls of advertising these days, and few have both. 

Fewer still share their findings, though this is changing as thought leadership becomes a bigger pillar of the e-commerce community. 

Luckily, we’re out there collecting the best investigations from the best operators and curious autodidacts that populate the DTC landscape. 

This week:

• All ads lead to Amazon - Investigating the halo effect 

• How much iteration do your ads need? - Should you be making minor edits or big changes to your ads?

• What that 8-figure brand does (that you probably don’t) - How you may need to evolve if you are pushing to break $10M

All roads (ads) lead to Amazon

Olivia Kory is one of these rare DTC scientists we mentioned. 

She’s building Haus.io, an attribution platform/agency that uses the scientific method to focus on nailing down incrementality, optimal ad spend, and omnichannel measurement. 

We’ve featured their findings in previous newsletters. Notably, Olivia’s take on whether you should even worry about incrementality in the first place (issue #19 | Finding Profit in Chaos).

Here’s their latest insight: 97% of ads show positive lift on Amazon. 

And that’s…

1. Regardless of the channel and format

2. Even if the ads are optimized for DTC sales (which they mostly were.)

This is the halo effect – when brand visibility in one place drives sales in another, unrelated place. For instance: if your presence in a retail store like Walmart drives sales online. 

Now, how strong was the halo effect exactly? 

Here’s what their results tell us: 

•  “16% of experiments showed a greater sales lift on Amazon than DTC”, and in Olivia’s words, “we’re talking about more than just pennies on the dollar”. This occurred more for brands with strong presences on Amazon.

•  “83% of experiments drove over a 10% halo effect on Amazon sales”. In other words, there is a high floor on potential results. Most of your campaigns will drive at least a 10% halo effect on Amazon. 

• “Social and video channels outshine search halo effects by 700%”. Most performance marketers already know or suspect that search campaigns (branded, non-branded, PMax) don’t provide as much lift as other channels, and Olivia confirms this theory with her findings. 

Social channels like Meta, TikTok, Snapchat and Pinterest drove a 7.9x greater halo effect on Amazon-sales-relative-to-DTC than search did, while video channels like YouTube and CTV hit 8.3x.

With Amazon Prime Day around the bend, Olivia also shared a few tips to make the most out of the insights above. 

Tip #1: Her first piece of advice (and one that is ALWAYS true when taking advice regarding paid media): run your own tests. 

• In Olivia’s cohort, some brands saw a 168% halo effect on Amazon sales from their Meta Reach campaigns, while others saw just 29%. 

• Many factors influence the outcome, but your portion of sales coming from Amazon and usual media tactics are two big ones. No one can tell where you’ll land, so test. 

• Extra tip: Run some small-scale tests during this Prime day to prepare for the Prime Big Deals in October. Haus has a feature that allows you to measure how awareness campaigns prior to the sale drive lift once the sale hits. 

Tip #2: Once you’ve identified which channels drive the greatest halo effect on Amazon, shift your budget towards these before sale day. Most likely a combination of social and video channels. 

Tip #3: Make ads that advertise the Amazon sale. Beyond increasing spend on strong halo effect channels, let them know that it's going down on Amazon as well. These will work particularly well in awareness campaigns on channels like Meta or YouTube where you won’t have to pay for an on-site conversion. 

Takeaway: However you advertise for DTC, it generally drives lift on Amazon. In fact, the more you rely on Amazon, the more lift it drives. 

Which probably means you should (i) get on Amazon if you aren’t, and (ii) not be scared to increase your activity there. Olivia’s results also suggest you can cut back on spend on intra-Amazon spend if it’s not working out for you and concentrate on generating lift through other channels. 

How much iteration do your ads need? 

This tweet kicked off a bit of a discussion, with established DTC marketers and operators pushing back and forth.

Soooo…Alex Cooper, who’s building Adcrate, went and tested the hypothesis on his agency’s ads. 

He looked at exactly 124 iterations across 17 ad accounts, 11 industries, and $12.4M in spend to see if the degree of iteration affected performance. 

To do so, he scored iterations by the degree of change they brought to the ad, with things like changing the hook scoring a “2”, versus changing the video creative asset scoring an “8”. Then he measured performance by spend. 

Now, there are a few weaknesses with this experiment, including: 

• Subjectivity of the iteration change score

• Measuring performance by spend

• Sample bias

Still, it’s a good first look at the problem. 

His results: a 0.42 correlation between iteration score and performance lift.

Fairly significant, though this does indicate that there are other factors at play. 

Still, it was clear from his data that more complex iterations result in higher lift. Conversely, small iterations result in a smaller lift % change. 

Takeaway: on an ad with high spend, say $500k, a small iteration resulting in a 10% jump in performance might be worth it.On an ad with low spend, the return won’t be so good. Either way, looking for a big change? Go for the big swing.

What that 8-figure brand do (that you don’t)

Peter Quadrel, founder of Odylic Media shares the 8-figure brand MO, which we thought raised some good points. 

• Forecasting all metrics daily on a per-channel basis. 

If anything, this demonstrates a closer attention to detail for each knock-on effect your marketing decisions have. It also helps you have a short and long-term roadmap for improving ad performance optimization and other things like inventory.

In Peter’s words: “Forecast or be lost”.

Check out his Cost Calculator and Spend Forecast tool here. 

•   Creative testing volume

Any digital marketer knows that only a few creatives ever become “winners”. Often, less than 10% of an account's assets drive 70% to 80% of its total revenue. 

As a result, great performance marketers have reached the conclusion that despite their name, successful creatives have more to do with quantity than quality. 

According to Peter, the fastest-growing brands launch at least 10 new ad angles a week, if not many more. 

Taylor Holiday from The Common Thread Collective agrees. 

Once you have enabled volume, the question remains on how to test creatives. 

For this, there are two principal schools of thought, which have clashed before (see how in our issue #17 - Creative Testing Showdown).

• The Big 2

Peter explains that 8-figure brands' main concerns are product and offer. 

Regular product tweaks are great for customer retention, and when made a priority, can also result in more operational efficiency and lower costs. 

Offer experiments, on the other hand, have more impact than any targeting or creative tweak could ever. 

So stop focusing on small hacks and tweaks, and focus on the more impactful stuff. 

• Whitelisting (aka UGC)

Running your ads through creators, blogs and theme accounts can significantly reduce blended CPA, thanks to a bunch of factors that we break down in our issue #11 - Keeping it Real. 

• Merchandising

Do you just advertise for all your products? Or rather, do you promote an astute selection? 

You do? Great. 

Based on what? 

Big brands select products that:

1. Are “new customer” profitable

2. Generate the highest 60-180 LTV

3. Liquidate sitting inventory (if cash is locked up)

4. Optimize for the current cash position

5. Leverage seasonality and market trends

According to Peter, you should be advertising about 5 hero products. Not much more.

• Use incrementality to measure ad performance.

Traditional attribution methods are imperfect. 

We’ll credit an interaction for a purchase, when in truth, a number of interactions, some difficult to pick up digitally, may have led to the purchase.

For instance, someone may have clicked an ad, not bought the product, and then seen a billboard on the street, which led them to go back home and buy. 

Incrementality is measured by exposing one group to an ad while keeping another (holdout) group from seeing it. The difference in conversions is an indication of the ad’s impact (though not exact unless both groups consist of identical twins experiencing the same life in a Truman Show-type scenario for the duration of the experiment). 

Shifting your focus to incrementality should help you take control of your ad performance and spend. 

Takeaway: Make sure your brand is monitoring the right data points, of which there are many, but keep your decision-makers focused on the big picture. As you scale, incrementality is truth (at least relative to traditional attribution), and creative volume matters more than quality. 

Godspeed. 

Quick hits

• Samay Devraj just launched Typecast, a human-curated library of the top 1% performing short-form videos.

• Mike Cessario’s (Liquid Death) 8 principles for success (credit: Jeremy Kuo)

• Bryan Porter’s (Simple Modern co-founder) 24 learnings & guidelines for success.

• Report - DTC EBIT margins are meaningfully below wholesale.