A sudden surge in website traffic is putting your live chat agents through the wringer. Luckily, you’re nothing short of solutions and decide to invest in an AI Chatbot to help manage workload. During your search, you come across multiple ad campaigns, read up on a few G2 reviews, and visit the websites of several chatbot services before finally making an informed purchase decision. But how does all this tie into the chatbot service earning better demand generation off your customer journey? And what does this have to do with marketing attribution? To explain, let's contextualize what attribution is.
An industry-wide shift towards digital transformation, customer centricity, and revenue accountability have rendered data-driven marketing more prevalent than ever before. As the number of channels available to a marketer continue to expand, the B2B customer journey has become far more scattered, convoluted, and nonlinear. As a result, there has been a rising demand for an organised, value driven MO that can help marketers consolidate their efforts and derive an actionable overview of their customer journeys. Enter: Marketing Attribution.
Marketing Attribution is a method of determining how your efforts — ad campaigns, website content, offline events etc. — have contributed to revenue, pipeline, and ROI. This is done by collating data across every customer touch-points and assigning attribution credits that provide a specific value to every interaction. For example, if a Facebook ad, Webinar, and Blog are three touch points in a customer’s journey before they make their purchase, attribution helps assign weights to each one of these points so as to provide an overview of the extent of their influence on the purchase decision.
Before delving into how some of the most popular attribution models work, it’s worth understanding the mechanics of attribution modelling. A broad compartmentalisation of attribution models would consist of rule-based attribution models and data-driven attribution models.
Rule based attribution models are models which use predetermined procedures of assigning attribution credits to touchpoints. The models used in this system are determined by various factors that are relevant to the brand and/or product. These factors include sales cycle, number of channels, opportunity cost of channels, etc.
Data-driven attribution models assign attribution credits to touchpoints based on premeditated metrics by assessing KPIs (key performance indicators). They also employ algorithms like keyword searches, touchpoints mapping etc. While neither approach may produce sure-fire results on representing a customer’s conversion journey, a combination of both these models may be administered to produce effective results.
Single-touch attribution models, like the name suggests, assign 100% of the attribution credits to only one of the touch-points encountered by a customer in their conversion journey. Some of the most common models include:
In this model, your customer's first touch-point — whether that be an ad campaign impression, content interaction, form-fill, or anything else in-between — is deemed the most important touchpoint in their journey. Hence, this interaction is assigned 100% of the attribution credit. For instance, let’s assume that you’re in the market for a project management software and come across an advert for one that catches your attention. The ad prompts you to visit the company’s website. After landing on their “features” page, you follow through with more research and come across the company’s weekly blog — before finally signing up for a demo. In this case, the advert you clicked on is your first impression of the brand and product. Hence, a first-touch attribution model, would reward the advert with 100% of the attribution credits.
While one might argue that prioritizing top-of-the-funnel marketing is a presumptuous move, keep in mind that a company’s attribution goal need not be a conversion in the form of a purchase. Instead, it could be something preliminary like getting a potential customer to create an account on your website. In such cases, single-touch attribution models may be most appropriate.
In a similar vein with first-touch attribution, a last-touch attribution model assigns 100% of the attribution credits to the touchpoint closest to a customer’s decision to convert. This would imply that the last impression made on the customer before their decision to convert was the most prominent in their journey. Continuing with the previous example of the PMS, the blog piece you come across before scheduling a demo would be the last touch. And so, out of all the touch points that influenced your decisions to sign up for a demo, attribution credits will be solely assigned to the final one.
Many businesses and marketing aficionados are of the opinion that single-touch attribution is not an applicable model on its own. It is often considered to be a one dimensional approach that fails to faithfully represent a customer’s conversion journey down the funnel. While single-touch models may have their own relevant use cases (like for products with shorter sales cycles), it may not be as effective in identifying the most influential touch-point in a B2B customer journey. If big data in marketing has proved anything, it's that customer journeys can be non-linear, sophisticated paths spanning over several channels and mediums. Having one consolidated attribute vouching for most of their decisions will rarely be sufficient. So while single-touch models have their place in marketing analytics, many regard it to be a diminishing model.
Multi-touch attribution modelling is the holy grail of marketing attribution. As customers’ buying patterns evolve and become increasingly scattered, a model that can track and account for all these interactions is of great demand in the marketing sphere. A multi-touch attribution model accounts for all the touchpoints encountered in a customer’s conversion journey. It’s a holistic view that helps paint an infinitely better picture of patterns and behaviour.
Remember to keep in mind that the goal of multi-touch attribution isn’t to just map out customers’ interactions. It is also employed to understand which touchpoints influence a customer the most, which touchpoints work in conjunction with each other, and what the relative probabilities of channel interactions among different customer paths are. With this established, there is still the issue of assigning credits to a now substantial number of touchpoints. To help illustrate multi-touch attribution better, here are a few of the most commonly used models:
A linear attribution model assigns attribution credits evenly among all touchpoints. While this model is far more illustrative than any of our single-touch attribution options, it's a relatively simplistic approach when compared to its nonlinear variants. Let’s assume that the total number of touchpoints in our PMS example is four: An advert, a blog, a review, and a retargeting campaign. Linear attribution would reward 25% of attribution credits to each of these touchpoint. Of course, in reality, the number of touchpoints a B2B customer goes through is significantly higher — so the weights for each one are likely to be far smaller.
The U-shaped model assigns attribution credits to all touchpoints — but assigns higher credits specifically to the first and last touchpoints. This would imply that your customer’s first and last impression are the two most valuable touch-points in their journey. Consider the same four touch points as with the previous example (Ad, Blog, Review, and Retargeting campaign). This time, maybe 40% of the credits will be assigned to the first and last touch points each. The two touchpoints in-between will receive only 10% each as they are deemed less influential to the conversion decision. The model laid out in a bar graph takes the shape of the letter ‘U’, hence the name.
Time decay attribution assigns attribution credits in an ascending cascade. What this means is that each touchpoint is given progressively higher credit, with the first touchpoint having the least credit and the last touchpoint having the most. This is an effective tool in mapping out a customer’s conversion journey. The model works on the implication that a customer’s need to convert becomes more prominent with each touchpoint — suggesting that the last one has the biggest impact. Again, using our handy four touchpoint PMS example, a time decay model would assign attribution credits in this manner: 5% for the advert, 15% for the blog, 20% for the reviews page, and 60% for the retargeting campaign.
In the end, a lot of the use cases of these models are subjective. The decision to opt for a specific model can be based on several reasons spanning from the nature of your product to the extent of your brand equity. More often than not, you will find yourself using more than just one model with several stipulations and custom values for each variant. Fortunately, the progressive ingenuity of AI and constant innovations around attribution modelling will render your experience less of a trial by fire and more of an intuitive, insightful practice.
Get the latest best practices in Marketing Analytics
delivered to your inbox. You don't want to miss this!!