.svg)
1. Switching from Last-Click Attribution to another static Model fixes your Marketing Attribution Problems
Many advertisers and digital marketing manager are very much aware of the fallacies of last-click attribution.
The seemingly simplest way to switch from last-click attribution, and especially if you use Google Analytics, is to choose one of the other static attribution models, for example, first-click, time-decay, linear or position based. The problem is, that these models are all arbitrary and more or less chosen based on marketing intuition. They are not based on reliable data analysis.
Working with static attribution models can have detrimental consequences on an advertiser’s marketing performance. Therefore marketing managers wanting to move away from last-click attribution should aim at using data-driven, algorithmic attribution models.
2. You have to spend more than USD 1000 per month to do algorithmic Attribution Modelling
The majority of marketing attribution solutions aim at large enterprises and require digital marketing budgets north of USD 5 million. These solutions cost more than USD 100,000 per year and are very people-intensive, meaning they have a group of sales managers and technical consultants involved in any particular project.
Usually these costly solutions have many different features included, e.g. marketing mix modeling, media planning and scenario analysis as well as evaluating offline campaigns (e.g. radio). These are all relevant tools for large advertisers, but not necessarily for startups or small and medium enterprises that want to move quickly.
The solution we created at Adtriba starts with a USD 200 monthly subscription fee and can be integrated literally within minutes. We built a solution to automate the data science processes within the algorithmic attribution modeling. Through this automation and the standardization of our solution, we allow our customers to become self-sufficient. We see ourselves as a technology provider and not as a digital marketing consultants.
3. Attribution Modelling is only for big Enterprises with many different marketing channels, not for Startups and SMEs
Similar to the before mentioned misunderstanding regarding the costs of marketing attribution solutions, this flawed assumption roots in how the existing solution position themselves in the market and talk about marketing attribution. Of course, the ROI of implementing a marketing attribution solution heavily depends on the cost of the tool relatively to the absolute amount of increases in marketing profitability, which in turn depends on the absolute size of marketing budget. So if a tool is very cost-efficient, that in turn allows advertisers with small to medium size budget suddenly to apply algorithmic attribution modeling as well.
Regarding the number of different channels (which usually correlates strongly with the size of the marketing budget) Adtriba even has clients that only work with Google AdWords and want to optimize budget allocation between different AdWords campaigns. Even in a minimalistic setting as this, our attribution systems is able to produce valuable insights for budget optimization within this single channel.
4. Algorithmic Attribution Modelling is a data science black box, you need a PhD to understand it
Some of the above-mentioned marketing attribution system vendors sell their attribution modeling algorithms as something close to rocket science and as a complete data science black box. This doesn’t necessarily lead to potential users of these systems trusting in the solution. Studies have shown that it’s essential that the users of marketing attribution systems trust the system and understand the basics of how the algorithms work. It doesn’t make sense to switch from last-click attribution, with its obvious flaws, to a data science black box, based on algorithms which none of the users understands.
That’s why we at Adtriba are completely transparent about our algorithms. You can read about the details here, but in essence we are using logistic regression to build a robust conversion model which we then use to calculate a conversion probability after every step in customers journey. The differences in conversion probability are what is being attributed to the respective marketing step. Additionally, we are using an optimized version of the Shapley value algorithm for evaluation and validation. We also produce and communicate a standard data science metric called AUC that helps our customers to understand how good the quality of the attribution models are.
5. Buying and implementing an attribution system is all you need to do to fix your marketing attribution
Implementing algorithmic attribution modelling to create a cross-channel perspective on all marketing activities is step one. But it is essential to implement new and change existing marketing processes such as budget allocation decisions and media planning.
The cross-channel attribution leads to understanding the performance of every campaign in a cross-channel context, i.e. the performance against every other channel and campaign. That is why there shouldn’t be fixed budgets per channel as this would prohibit dynamically relocating budgets between channels and campaigns.
6. Retargeting always loses to other channels when abandoning last-click attribution
It’s not difficult to understand, that the rise of retargeting as a marketing channel heavily profited from last-click attribution. Retargeting per-se is a lower-funnel marketing tool and hence depends on other traffic sources bringing in the visitors that should be retargeted.
Since its amazing increase in popularity retargeting services have developed quite a bit. With most of the retargeting services, you can choose which part of the funnel you want to re-target and frequency-caps are in place for most advertisers.
That’s why it’s not surprising that we see many retargeting campaigns at our customers that are undervalued through last-click. Our algorithmic attribution model would suggest spending more in retargeting compared to last-click attribution.
Generally speaking one must be very careful with generalizations such as “retargeting/affiliates campaigns lose when abandoning last-click”, which make for a good sales-pitch but don’t reflect complex marketing realities.