Precisely attributing marketing spend is a top priority for marketers. Now is the time for organisations to identify attribution solutions that provide a single view of the customer’s journey, allowing more informed predictive modelling and data-driven audience segmentation.
While there has been a significant step-forward in how brands are measuring attribution by adopting a more data-driven approach, their accuracy is heavily influenced by the choice of modelling technique; make the wrong assumptions or adopt the wrong technique and you risk ending up with a solution as inaccurate as the simplistic rules-based offerings.
To help you understand what to look out for, we’ve pulled together a list of the top seven features that will provide you with the most accurate approach to attribution.
7 things to look for in an attribution solution
When it comes to attribution solutions, here are the major factors, to determine what you need, and the types of features to look for. Jaywing’s sophisticated approach to data-driven attribution is highly sophisticated and includes all of these features.
- Data-driven modelling technique. Many advanced solutions, like GA 360, use the Shapley Value to attribute spend. While this is a great model, we recommend a combination of the Shapley Value and Random Forests models to precisely estimate the probability of success. This is down to the fact that the most accurate models recognise the contribution and synergies each individual touchpoint brings and understands as much of the detail of the journey as possible – a simple model generally won’t suffice. The Random Forest modelling technique we adopt means: all events in the journey are considered, not just the last touchpoint; all journeys are included, both successful and unsuccessful; the timing of events, and time between events is contemplated; the detail of the touchpoints, down to keyword level if necessary is part of the information considered by the model. And finally, the model must generate fair, unbiased values that represent the importance of each touchpoint: the Shapley value technique does this, albeit with serious computational demands. Without access to considerable computing power, Shapley calculations become very difficult and slow. Jaywing has invested heavily in efficient coding and cloud-based computations so that this isn’t a problem.
- Custom data sources. Integration with 3rd party data can be complicated with GA 360. Although Google provides some tools to ingest 3rd party data, it is often not an easy process and requires specialist setup. As data specialists first and foremost, we start by combining all your data in one place. CRM data for example, is essential to include to know who your customers are and what actions they’ve taken previously.
- Customer journey tracking. Solutions like GA 360 can have significant limitations, looking at only four touchpoints, and often in up to a maximum of a 90-day period. This places significant limitations on tracking individual customer journeys, which means you may not see the entire cross-channel customer journey. For example, where your brand has longer customer journeys, i.e. when a purchase is more considered, it can mean important interactions with your brand and marketing early in the journey will be lost with only the last four touchpoints being measured. This will undervalue upper-funnel marketing and overvalue the last few actions. Additionally, when customers purchase multiple times, if the second purchase is outside of the 90-day period a customer may look like a new prospect. That’s why we vary the length to suit your requirements, tracking as much of the customer journey as is appropriate. Data is being used to make decisions and people need to trust that data, so it’s essential that you’re tracking the entire customer journey and every touchpoint
- Display impression data. With the phasing out of third-party cookies across popular web browsers, it’s going to become more and more essential that you can model display impressions. This will be a significant loss to a lot of attribution providers. While it may be possible for them to provide some display impression data for activity run through the Google stack, a sophisticated custom-built display impression model will provide greater accuracy to ensure any display impressions are still measured and valued as part of the customer journey.
- Base sales. Not all sales values are attributed to marketing channels. A baseline is the attributed value of everything outside of the marketing campaign being measured — or in simpler terms, what would have happened anyway. By modelling your baseline you can understand the incremental impact of your marketing and explain and account for fluctuations in performance due to external factors. With an accurate baseline you can uncover new insights about each of your marketing channels well beyond determining the sources of conversions.
- Cookie compliance and opt-in. Many clients are looking to implement cookie acceptance solutions based on legal advice, which prevents any marketing and analytical cookies from being set until the user has voluntarily opted in. The resultant data collected is significantly reduced in volume and includes a bias (brand advocates are much more likely to accept such a solution). It is important to take this sample and adjust to represent the true impact of data collected, modelling the gaps and adjusting for any bias. Our data science team has experience adjusting for this issue, an area where tools such as GA 360, where a one-size-fits-all solution is necessary, struggle.
- Customisation. Return on investment is what we all seek and simply implementing a new analytics tool is not going to drive a return; it takes people, knowledge, processes, and determination. Every brand we talk to needs to get something different out of their attribution solution, and therefore require a level of customisation to ensure the approach works for their brand and situation. Solutions like GA 360 are built to service a large number of clients rather than bespoke for a particular business or industry, and as a result, come with limitations on what is achievable. In addition to the tool, service, support and customisation is provided at an additional cost through Google Partners.
In summary
Most solutions fail to account for the entire customer journey, fail to capture complex interactions and apply rules that don’t fairly attribute spend across channels and touchpoints. Yet marketers need an approach that considers the impact of all possible inputs on behaviour, across the entire customer journey. This allows you to understand which campaigns, channels, and activities are working at which touchpoints along the customer journey, from initial awareness to desired actions, purchase, and advocacy.
GA 360 and the free version is great for many use cases, but for companies that need increases in data limits and advanced reporting and attribution, a more sophisticated approach may be the better choice. No matter which approach you select, it is generally best to use expert consultants to make sure that you are laying the foundational groundwork for an implementation that will meet your needs today, and also scale for your future needs.
Reveal nothing but the truth
At Jaywing everything we do is grounded in data science. We understand the maths and we’re careful that our solutions are ‘correct’ and will give a better, more accurate answer. We use data and analytics to drive performance in moments that matter. With the track record to prove it, every creative detail or technical tweak is motivated by optimising results. Based on facts, not assumptions.
Our approach to attribution is incredibly sophisticated, yet amazingly simple. We combine all your online and offline data, removes your base sales and use powerful machine learning techniques to attribute a fair share of value to sales. Revealing nothing but the truth.
Find out more about Jaywing’s unique approach to attribution or get in touch with us today.