Boost Marketing Impact With Smart Attribution Models

As a marketer, one of the biggest challenges is not knowing where your  marketing budget is having the biggest impact. It’s common to be overwhelmed by a lot of metrics and not knowing how to accurately measure performance and optimize spending.

The problem, which I discovered, was a focus on too many metrics instead of the one or two most important key performance indicators (KPIs). So, it's crucial to define the one or two KPIs that indicate marketing success specifically for your business goals and objectives.

For example, for an e-commerce company focused on growth, the core KPIs may include:

• Return on Ad Spend (ROAS): Calculates revenue driven from advertising spend
• Customer Acquisition Cost (CAC): Reflects cost to acquire each new customer
• Customer Lifetime Value (LTV): Revenue per customer over an entire relationship

A CAC of $100 may be viable for a $1,000 product but too high for a $50 product. In such a situation, a high CAC may be acceptable only for high LTV customers. Therefore, you need to define indicators of success based on your objectives.

With a tight focus on the 1-2 metrics that matter most, you can now start applying attribution models which can be very powerful. These models provide a way to methodically track how much credit to assign different marketing channels for driving your core KPIs.

Let’s look at how to implement attribution to take the guesswork out of marketing measurement.

1 Pick Attribution Models Aligned to KPIs

Properly attributing marketing performance and comparing channel impact is challenging but critically important. In my experience, the key challenge is ensuring that there is no “cannibalization” happening – meaning you clearly understand the unique value each channel provides in reaching your goals. 

Of course, it’s difficult to make attribution 100% transparent. But, selecting the right model and interpreting the results cautiously can help you greatly in clarifying a channel’s effectiveness.

So, what exactly are attribution models and how do they work? Attribution models are processes of assigning credit to different marketing channels and touchpoints involved in a customer’s journey towards conversion. There are many different types of attribution models, each with their own advantages and disadvantages.

For instance, there’s the first-click attribution model. In this model, the attribution is going to allocate all the credit of the purchase to the first touch point. A relevant use case for this is to understand how effective your top of funnel efforts are. This model basically gives more importance to your upper funnel where your customers saw your brand first.

Another example is the last click attribution model. This gives full credit at the end of the journey to the final touchpoint before conversion. The advantage is isolating the channel most directly tied to driving sales, but it fails to capture the influence of multiple upstream touchpoints.

There are many other models as well, such as linear which attributes equal weight along the entire customer journey, time decay which weights touchpoints by recency, U-shaped which gives credit to first and last channels only, and more. There’s also marketing mix modeling that collects historical data about when the campaigns were running, how much budget was spent, and how many conversions took place.

Given the variety of models, to understand which model is more suitable for your situation and business, I would recommend asking yourself these questions:

• What are your objectives?

• What resources do you have available?

• What stage of growth is your company in?

• What type of business and target audiences are you focused on?

For example, if your objective at the moment is a bit more upper funnel, then it's about creating awareness and growing traffic. For this goal, you should prioritize the first-touch attribution model. If you have access to a robust data collection tool, then multi-touch attribution could be the best choice here.

In fact, at one of my previous companies, we switched from a multi-touch attribution model to a last touch attribution model. This switch revealed to us that the results were really skewed towards one of the channels, which was search engine paid ads. This was mainly because search ads are typically very demand-driven and they're a lower or later stage kind of a channel. Later on, this helped us adjust our investments in search engine ads accordingly.

Ultimately, the goal here is for you to understand the effectiveness of different marketing efforts and to allocate marketing resources appropriately. All of the attribution models have their advantages and disadvantages. Hence, it's best to select one that works best for your business and also to interpret it with full understanding of its potential drawbacks.

2 Avoid Common Attribution Mistakes

When it comes to marketing attribution, I’ve seen several common mistakes made by professionals.

The most frequent mistake is relying solely on one attribution model without carefully interpreting the results. It’s important to analyze performance through multiple models to reduce the risk of misinterpretation or missing key insights. For resource-constrained teams that may not be able to build complex attribution, even basic incremental testing can help you further clarify the impact of your channels.

A second dangerous mistake is depending only on platform-specific attribution from the likes of Google and Meta. These can be inherently biased to showcase better performance for their associated channels. Therefore, I’d highly recommend supplementing with unbiased 3rd party attribution tools like Branch or Looker etc.

Another common pitfall is using industry benchmark comparisons without context - i.e. focusing only on direct marketing costs rather than the total cost.

For example, if an advertising campaign achieves an ROI of one, that wouldn’t really be a breakeven campaign because there are more costs to your campaign than just the marketing dollar spent. For instance, there’s the cost of marketing resources. There are the final steps like customer experience and customer support. In addition, using customer lifetime value to set targets may also not be a great idea because you’re making a bet that historical behavior will continue in the future, which might not be the case. 

Keeping your attribution model static is also another common mistake. This is really because your business keeps evolving. And as I mentioned, your business stage is an important factor to take into account when selecting attribution models. And of course, you can go through different growth stages during your time at the company.

Therefore the core of avoiding these mistakes comes down to a commitment to continually question performance with a critical eye. Are multiple models showing similar results? Are we confident in the objectivity of the analysis? Have targets been set reasonably based on full business context? So, progress in this realm requires recognizing analytics gaps and proactively seeking answers through expanded attribution capabilities, incremental testing, and ultimately sound interpretation.

1  Leverage Multiple Models for Holistic Analysis

Rather than relying on a single attribution model, leverage insights from at least two models aligned to your goals. Carefully evaluating results through different attribution lenses reduces misinterpretation risks and provides more holistic visibility into performance across touchpoints. Frequently reassess selected models as objectives and growth evolves.

2  Interpret Platform Attribution Cautiously

Exercise caution when utilizing attribution analysis from marketing platforms like Google and Meta. While convenient, these have inherent biases that may overstate performance for associated channels vs others. For balanced insights, supplement with neutral 3rd party attribution tools (e.g. Branch, Looker etc.) that can provide an unfiltered view of influence across all channels.

3  Set Attribution Models Against Business Context

Ensure attribution model relevance against factors like overall objectives, growth stage, and resources. Avoid common pitfalls like mirroring benchmarks without customization, focusing solely on direct marketing costs or historical customer lifetime value when setting targets. Contextualize your model performance against your unique business situation for optimal visibility.