eCommerce Mission Control

eCommerce Insights from Astronaut Party

Hello,

This week we’re talking tools. Specifically, we’re sharing why we think multi-touch attribution is essential for digital marketers and the AI tools we’ve been playing with.

As always, feel free to reply with feedback.

Up and to the right,

Luke & The Astronaut Party Team

The Biggest Lessons from Multi-Touch Attribution & Why You Need It

By Carlo Burriesci

META and Google are happy to report on their own performance, but should you trust them?

We are lucky that we don’t have to. Today, there are several third-party platforms that track each user's interactions with our brands and help us analyze how our media influenced those interactions.

Why is this important? Because we understand that users can interact with a brand many times, and across many platforms, before making a purchase. If a user clicks on an ad, the platform running said ad will want to take all of the credit. In other words, in-platform metrics will almost always inflate their performance.

Over the past year, we’ve moved our clients to Northbeam for their multi-touch attribution tool to help with this analysis and tracking.

Here are our biggest learnings:

1. The creative that gets reach isn’t necessarily the creative that converts.

Northbeam’s creative analytics tools show us, under a multi-touch model, which creative is truly driving performance. Not surprisingly, successful ads don’t necessarily drive one whole purchase but rather play a role in driving many purchases. In-platform metrics, on the other hand, assign full credit to a clicked ad. The result? An in-platform algorithm that optimizes toward a couple of lucky ads clicked at the right time by a user in their purchasing journey. The impression might be delivered at the right time, but it doesn’t mean that is the ad that caused a user to convert.

2. Greater budget control is better than platform automation. 

Platforms like META are pushing ad buyers to leverage automated budgets to efficiently allocate spend across various ad sets under a single campaign. Northbeam, however, regularly shows different top-performing audiences compared to what’s shown in the platform, leaving our budgets prone to wasteful spending as the algorithms optimize toward the wrong audiences. We’re not afraid to control budgets at the ad set level to manually support the audiences Northbeam tells us are the true performers.

3. Top of Funnel vs Bottom of funnel differences are huge. 

Differences in ad and audience performance inevitably lead to large differences in our overall campaign performances. In-platform metrics were showing that our prospecting campaigns were hardly delivering a positive ROAS, while our remarketing campaigns were blowing it out of the park. Northbeam showed our prospecting campaigns were, in reality, much stronger while retargeting, though strong, wasn’t exceeding expectations as much as we initially thought. Without this information, we’d have pulled the plug on some effective prospecting audiences and beefed up retargeting budgets more than necessary.

The big takeaway? 

As much as platforms praise an ever-growing “hands-off” approach to media buying, this only works if companies trust in-platform metrics. To answer our earlier question, no companies shouldn’t trust platforms, and should instead turn to a more sophisticated attribution model like multi-touch attribution analysis. As a result, this means manual maintenance is still essential to get the best performance from ads.

AI Tools for ECommerce

By Beth Teutschmann

It's no secret that AI is affecting nearly every industry. And those who don't adapt will be replaced (probably by AI, if we're being honest). 

We're regularly trying out new AI tools to see what fits into - and improves - our workflow. 

Here are some of our favorites and some that we're testing right now:

1. AI for Product Photos (like Photoshop’s Generative Fill). 

We've used this to create product backgrounds for organic social posts. It's not perfect (and we definitely don't recommend asking it to include a person because you can't unsee the hellspawn it's going to generate), but it can save a lot of time from having to generate a background by hand. We've found that most backgrounds need a little tweaking, but it's still a time-saver when compared to hunting down the right stock image.

Speaking of stock images, they're saving us money. We don't have to buy digital backgrounds anymore; we just ask AI to make them for us. As long as you have someone on your team who knows basic photo manipulation, you're golden. 

2. AI for Ad Audiences. 

iOS 14 was a death knell for third-party cookies. Since then, the marketing world has been forced to get creative to figure out how to reach the right users. AI has helped determine which characteristics matter for conversion and then target more similar users. A number of providers offer services that track your users and use AI modeling to generate new audiences, like Black Crow and Proxima.

3. AI for Meeting Recaps & Notetaking. 

Video call note-taking services like Fireflies.ai can not only create a transcript of your video meetings but also give you a recap. This is really helpful if you have team members who can't make it to a critical meeting. They can get a quick summary of who said what and a full transcript if they need more details. There is no need to ask someone in the group to take and/or transcribe notes; everyone can relax and participate in real time.

4. AI for Custom Workflows. 

AI reached new heights in November of 2022 when ChatGPT was unleashed upon the world. By January 2024, OpenAI released a marketplace for custom ChatGPT apps

And the world of AI has continued to evolve at record speed ever since. 

One of the newest tools we're testing is YouAI from MindStudio. YouAI simplifies the AI building process so you can build your own customized AI without knowing how to code. 

For example: 

You can give YouAI your client's brand guidelines and have it create customized marketing and sales materials, landing pages, etc. 

Or you could upload your current digital campaigns and use it to help you optimize your campaign budget. There's also a lot of integrations to choose from, like MailChimp, Slack, and more. 

You also have the option of choosing which AI model to build your outputs on - ChatGPT, Gemini, Meta AI, etc. No more switching between web pages, you can configure your AI to give you an output in multiple large language models. 

We're currently testing out YouAI by building an AI to help us write newsletter content for our owned brand, Code&Quill

We're feeding it content from past newsletters, brand guidelines, and relevant articles and giving it a suggested tone and word count.

The final product isn't 100% perfect, but it's much closer (and more streamlined, once the AI is built) than just using ChatGPT. 

Next, we will test its use for ad copy generation by feeding it client brand guidelines and past winning ad copy. This should be especially useful for quickly developing copy variations with character limits (we're looking at you, Google Ads).

5. AI for UGC Ad Creative. 

Writing scripts and working directly with influencers is complex, slow, and expensive. Now, there are services like CreaterKit that have developed models based on UGC content and, with a bit of coaching, can create full-fledged influencer-style ads from AI. Check out an example from Code&Quill here. Maybe it's still a little too creepy? Perhaps it’ll get even more natural in a couple of months.

We’re excited to see where AI takes us. So far, it’s helping us save a lot of time on our day-to-day agency tasks, which makes us more efficient and effective overall - something our clients greatly appreciate.

Joke of the Week

What do astronauts listen to on the radio?

Nep-tunes. 🌝 🎵