Why I don’t bother with attribution anymore
The first marketing job I had was at monday.com. I worked there almost two years, and to a certain extent this period shaped me as a marketer.
In many companies, marketing is seen as fluff, focused on pretty visuals and nice-sounding copy. But at monday.com, it's all about revenue generation. You were always expected to know the numbers: How much revenue did you bring in? What's the ROI? What's the CAC?
Today, I still take this approach. When I start a conversation with a new client, I usually talk about financials: What is your ACV? What are your CAC targets? To reach your goals, what revenue do you need to generate in the next two quarters?
Their ICP and existing campaigns are important, but they are only means to an end, which is revenue.
Yet, one thing I do not take away from monday.com is how they handled attribution. At monday.com, attribution was king: they built complex attribution models that were supposed to make financial analysis possible at both the campaign and ad levels.
The models attributed every customer to a specific campaign that supposedly generated it, so campaign managers always tracked progress through top-line metrics: campaign X generated $25k in July, campaign Y generated double the ARR but spent 50% more, etc.
This is something I no longer do.
The first reason is that attribution is becoming increasingly difficult. With privacy restrictions, you cannot track every customer interaction. For example, in 2019, you could tell if someone saw an ad on Linkedin without clicking on it, but now it's impossible.
(Also, even in 2019, some things weren’t trackable. For example, someone viewing YouTube ad and telling a colleague about it.)
Second, because I don't need to. monday.com was a huge company who spent millions on ads. I, on the other hand, mostly work with early stage startups, and early stage startups don't need fancy attribution models. For them, correlation is causation:
When you launch a new channel with a significant budget, and you see a significant uptick in your key metrics, this means that the channel is working. When you do not see such an uptick, that means that the channel is not working.
When you’re early-stage, that's all you need for attribution.
In-platform metrics
The only problem with this approach is that it only provides you with insight into the channels, not into specific campaigns.
Let's say you're an early-stage startup that launched a YouTube campaign for $12K/month and saw a 50% increase in product-qualified leads. Having said that, you can now say with confidence that YouTube is a great channel, but what specific YouTube campaigns perform the best?
To optimize your campaigns, you must answer this question.
For that, I rely on in-platform metrics, where the most important are in-platform conversions. In-platform conversions refer to the way the ad platforms attribute website conversions (like signups) to campaigns and ads they believe were responsible for generating them.
Don’t get me wrong: I would never rely on those metrics to decide whether the channel even works. However, once I've made that decision, based on top line metrics, I can use those internal signals to determine which specific campaigns or ads perform best.
Conversions should be tracked across your entire product funnel. For example, I tracked the following conversions in Datree:
Pageviews: Users who visited https://datree.io
Meaningful pageviews: Users who visited more than two pages at https://datree.io
Docs pageviews: Users who visited Datree’s documentation.
Signup: Users who signed up to the product.
Ideally, you should track deeper conversions, such as paying customers, product demos, or active users. However, most early-stage startups lack the volume to make that possible.
In addition to conversions, engagement can also reveal useful information.
Among marketers, engagement is usually considered a vanity metric, but I disagree: In my experience, engagement and performance are strongly correlated, so optimizing your campaign for engagement can be very effective.
(There is only one condition: You should discuss your product explicitly in all your assets. It is very easy to generate engagement with content that is not promotional, but that won't help you achieve your goals.)
The engagement metrics you should measure vary based on the channel and medium. If you're distributing videos, you should measure watch time: How many viewers watched 50% of the video? How many watched 75%?
This graph is from one of my weekly reports for Datree, showing how many users watched 75% of the videos we distribute, and how much we paid for those views:
This is specific to video ads, so for image and text-based ads, focus instead on likes, shares, and comments. Comments are especially interesting, since they reveal how your target audience is actually responding to your ads.
Traditional attribution
In terms of attribution, I'm a bit unorthodox. I don't do UTM tracking, I rarely check HubSpot about the source of my leads, and I’m not concerned about ad blockers. In one channel, however, I take a more traditional approach to attribution: Google Search.
First, because Google Search is an expensive marketing channel with limited brand awareness effects. If I'm not 100% sure I'm making money with it, I'd rather spend that budget on social media advertising.
Second, because Google Search is about the only channel where traditional attribution works:
As it is based on clicks, cookies can be used to track a user's activity.
Since it's an intent channel, converting from click to conversion should be relatively straightforward.
Consequently, I expect to see not only a positive trend in KPIs to indicate that Google Search is working, but also actual technology-based attribution (in most cases, an indication in HubSpot's Source column that Google Search was the first touchpoint).
The same is true for SEO.
I wouldn't be as obsessed with it as I am with Google Search, but as a rule of thumb: If you have 1000 visitors a month to your blog, but you can't find a single lead who visited the blog as the first page on your website, it could indicate your blog activity isn't converting.