Jun 19, 2018

Become a champion in attribution modeling: a how to guide

By Supermetrics

Marketing measurementWeb Analytics

[ Updated Jan 31, 2024 ]

11-MINUTE READ · By Tina Arnoldi

We carry a computer in our pockets. “Surfing the internet” isn’t something that’s done at a desktop computer. And we rarely get online at our desks and made an immediate decision. Instead we start – or finish – searches while on the go with our “pocket computers”. Since mobile is a primary channel for accessing the internet and we switch among our screens and marketing channels throughout the day, we need to account for fragmented behavior in our measurement plan. Not only is the device of choice fragmented, the channels we interact with before making a decision has grown.

Understanding how customers interact with brands every day is not straightforward since they switch devices and channels before making a decision. B2B decision makers especially are rarely going to make a decision with one visit at a desktop computer .

These non-linear journeys present a challenge for how to best optimize the full marketing spend. How much money should be invested in social media ads on a mobile device? How much should we put into the Google search ad right before conversion on a desktop? The example below can make a case for investing more into desktop marketing since it shows mobile conversions are more expensive.

 

But remember, this is data for only one business. The picture may look very different for other industries.

Get out of your silo

To choose the right attribution model for our business goals, we need to first get out of siloed thinking. We have departments for social media, search engine marketing, email marketing, and content marketing but these departments need to collaborate as part of the larger strategy. They all operate with a goal of bringing in new business but how do they work together? Attribution helps us understand how each contributes to the larger goal by assigning attribution credit. This is why we want to avoid last-click attribution because we miss out on all those touch points that happened before a conversion and often discount the true value of mobile .

With the right attribution model, we can connect the different touchpoints and credit them appropriately.

Conversions first

To use attribution modeling in AdWords, you need to first set up conversions in AdWords . (If conversions are already set up, it may be a good time to evaluate your existing AdWords conversions to ensure you look at the right metrics for your business. ) Once you track conversions, then you can select the appropriate attribution model. You can’t analyze whether what you do works until you first know what you are measuring.

Attribution modeling

“An attribution model is the rule, or set of rules, that determines how credit for sales and conversions is assigned to touchpoints in conversion paths. “ Remember that the right attribution model values all click touch points but each model is a little different. Choosing the right model for your goals is important because we optimize campaigns according to the model. With the wrong model, we do not allocate credit properly.

We want to measure activity across the full consumer journey because we can better optimize our campaigns when we understand which channels provide the best ROI. Assigning proper credit to each channel helps us make decisions about how to allocate our budget and better leverage our tactics.

The best Attribution Model, Data Driven Attribution (DDA) is not available to everyone. It requires 10,000 clicks and 400 conversion actions . It’s a great model though because it uses our account data with Google’s machine learning and drills into the contribution of each keyword across the path, comparing converting and non converting visitors. It is an improvement over the other rules based models because they apply the same rule throughout the customer journey, not accounting for your marketing strategy or seasonality. For advertisers who do not meet the criteria for DDA, the other options available are described briefly below.

Although I do not recommend the First Interaction/Click and Last Interaction/Click ( and neither does Google ) , the appeal is understandable The First Interaction initiated the journey to purchase so it is especially important for newer brands that need to drive awareness. The Last Interaction is when the purchase was completed. It’s valuable to know which channel it was, the message when the conversion happened, and the segment for that last click. But there was likely research in between those clicks that influenced the final decision

Linear is a more neutral model and is a good choice if you are not sure which one of the other rules model is best for you. The Linear model assigns equal weight to all channels so it does account for all touchpoints, but do all touchpoints truly have the same value? Time Decay gives more credit towards the end as the person gets closer to converting and Position-based puts the most emphasis towards the beginning and end of the journey.

Measuring performance

Like everything else in AdWords, although you may initially panic at a decline in performance after a change, don’t make decisions based on incomplete data. Use caution before making any changes reactively, including updating bids

When you apply a new attribution model, consider the average time lag in your conversion cycle and ignore recent days when looking at performance in AdWords. Allow enough time for conversions to happen, keeping lagging consumer behavior and purchase cycles in mind. Is your average purchase cycle two days or two months? If it seems that conversions decreased the week after applying a new model, give it another week. In other words, look at last week’s conversion data today , not today’s data.

AdWords search attribution

There are a couple ways to view performance based on your model. In AdWords, view it under Measurement > Search Attribution > Attribution Modeling. This allows you to compare models ( Last click and First click highlighted in yellow below) and select a history window up to 90 days. You can look high level at performance of your campaigns or drill into specific keywords. The % Change (on the right) indicates the change in conversions based on the two models you compared

Analytics model comparison tool

In Google Analytics, you can view the Model Comparison Tool (under Conversions > Attribution to view attribution across all channels, not just AdWords. Here too you can view the percent change in Conversions based on the model

Tracking your progress in Google Data Studio

Next, it is important to set up proper reporting in place in order to analyze data and make relevant adjustments to your strategy.

With Supermetrics Google Analytics Data Studio connector , you can see the path to conversions and where in the path each channel had an impact. In your report, you’ll see variation with some channels bearing more weight at the beginning (First interaction) and other channel interactions impacting conversions more at the end (Last interaction).

About Tina Arnoldi

TinaArnoldi

Tina Arnoldi is Analytics and AdWords Qualified and one of the few people in the United States recognized as a Google Developer Expert(GDE) for marketing. Her agency, 360 Internet Strategy, is also a Google Partner. You can learn more about her on LinkedIn

Stay in the loop with our newsletter

Be the first to hear about product updates and marketing data tips