November 20, 2017
We’ll save you from citing the infamous advertising attribution quote that all of these blog posts start out with. If you’re reading this, you probably already know how challenging attribution is–especially for mobile marketers.
This year advertisers really tackled this challenge, but the fact of the matter is….
- Advertisers are set to spend $77 billion a year on digital advertising.
- On average, marketers spend on 6 (B2C) to 8 (B2B) different advertising channels.
- In 2017, 77% of advertisers still aren’t confident in their attribution solution.
As direct response marketers, we’re committed to helping our clients allocate resources as efficiently as possible. In working with companies big and small across industries, we’ve run across several common attribution model mistakes. In this post, we’re addressing three of those challenges we ran into a lot this past year and how we are helping our clients overcome them.
Mistake #1: One-Size-Fits-All Attribution
Solution: Your channels and users are complex. Your attribution model should be too.
The major flaws in antiquated attribution models are that they don’t account for touchpoints across multiple channels or devices. Facebook themselves are championing this movement, encouraging advertisers to ditch the ‘last click’ mentality with their recent campaign. For apps, mobile marketing partners (MMPs) have risen to the occasion, piping in and de-duplicating conversions across channels and devices, so you have one source of truth.
The default last-click attribution model just doesn’t have a place in 2017, and yet today only 50% of marketers have implemented multi-touch attribution models. Of the other 50% who have, many are approaching them as one-size-fits-all. They choose one of the popular options…
In reality, each company should develop their own model based on their users’ behavior, campaigns, advertising spend, top performing channels, and nature of their product. Here is a simple framework we apply in helping clients develop a suitable attribution model:
1. Start by mapping your channels to the buyer’s journey.
What channels are often the first- vs. last-click? What channels have more view-throughs than click-throughs?
2. Weigh the nature of each of those channels to the place it plays in your marketing mix.
Is X channel better for conveying high-quality value props or quick takeaways? Does Y channel attract high or low LTV users?
3. If you have distinctly different user types (you probably do), note the differences in paths.
Does behavior differ regionally? Do users behave differently across your different offerings?
4. Weigh the average number of touchpoints to the period of time over which you’re spending.
Is your message time specific? Do users with long-term brand exposure tend to have a higher LTV? What is your ad frequency on each channel?
Regardless of what attribution model you implement, these learnings can be powerful in their own right. They can help you determine what kinds of creative you develop, or how much you’re bidding on each channel.
Mistake #2: Misattribution & Missed Opportunities
Solution: Challenge the standard attribution windows to reflect your buyers’ journeys.
Most media partners and acquisition channels have their own default attribution windows, the periods of time given between conversions and ad views or clicks.
These defaults can (and should) be adjusted. It falls on each organization to decide what’s important to their business, and what windows give the most realistic view. You should also think through the appropriate windows for app installs vs. in-app events vs. purchases.
What goes into deciding how to determine the appropriate attribution window?
When setting a benchmark for your attribution window, go beyond your intuition and look into historical data, user research and even what your peers and competitors are doing. Weigh this information along with the nature of your offering, price points, and general brand awareness.
The next way to determine the appropriate window is to analyze the impact of historical ads on conversion behavior, isolating the contribution of ads in different lookback windows.
Why is it so important to get your attribution windows right?
Having a too-long attribution window will capture more conversions for longer sales cycles, but will lead to attributing too much, praising media that didn’t indeed contribute to conversions. If your window is too short, you may be undercutting your purchase journey and not giving credit enough to upper-funnel activities that drive awareness.
Making either of those mistakes will result in losing money on misattribution, or the allocation of spend in the wrong places, missing opportunities for your business, and undercutting high-impact but low-attributed activities.
Mistake #3: Mistaking Attribution for Causality
Solution: Consider running an incrementality test to truly prove impact.
Once you’ve used your own data and hypotheses to inform your ‘Goldilocks’ attribution window, it’s important to remember that attribution does not equal causality.
Just because you can attribute a digital correlation between an ad view or click does not mean that the serving of that ad directly resulted in the conversion. The only way to determine what is actually impacting conversions is to run an incrementality test (aka lift tests).
Incrementality is the ultimate source of truth for measuring ad spend effectiveness. In short, it’s an A/B test wherein you show ads to a group of random people and ‘hold-out’ another group of random people. You then compare conversions and user behavior between the two groups of people to determine whether or not the group served ads did indeed convert more (or less).
Although developing a comprehensive marketing attribution model may help you spend smarter and sleep easier, it also requires A LOT of additional work and money. Determining how complex your attribution model should be is a process each company should make early and revisit often. In determining the best approach for your organization, we recommend that you constantly weigh those pros and cons, taking this basic approach…
1. Do the hard work first upfront
Ask these questions upfront and make sure your your boss and your boss’s boss care about the impact of your efforts.
2. Integrate with an MMP
This is a must for modern performance marketers to de-duplicate conversions and ensure you can trust your data.
3. Ditch your last-click model
Look to your users’ behavior across your channels to determine the best suited multi-touch attribution model.
4. Challenge your default attribution models
Use historical data, industry knowledge, and user behavior to set attribution window benchmarks across your channels.
5. Put your channels to the test
For the most advanced advertisers spending a lot on many different channels, run incrementality tests to determine the actual impact of your channels.
As tools and technology catch up, we’re glad that more advertisers are tackling these challenges and this approach is becoming more of the industry norm.
We design and execute highly scalable paid social programs for world-class brands in a mobile-first world.More About Us