Chapter 5 Flashcards
Why is the Shapely Values Method considered better than the click-based attribution method?
Because it overcomes one important limitation of the click-based attribution method: you do look at what happens if the channel is not there.
Shapely Values Method
Looking at comparable ‘paths to purchase’, with the difference being in that one channel is missing.
Converted customers
People who have completed a transaction.
They have visited the website, signed up and created an account, and purchased from the website.
Authenticated users
People who have signed up for an account, but no money has changed hands (no purchase yet)
Visitors
People who have been to the website but have not provided personally identifiable information necessary to sign up for an account.
Non-visitors
People who have never interacted with the firm’s website.
Four types of (potential) customers
- Non-visitors
- Visitors
- Authenticated users
- Converted users
What is the best way to measure if a channel is responsible for the sale?
By seeing what would have happened if we would have left the channel out.
Time travel and alternative universes are not an option.
BUT, experiments are an option. Through experiments, alternative universes are created.
How are attribution in marketing and attribution in sport comparable?
In both cases, the whole game changes when one channel or player is taken out. The whole dynamic changes.
Attribution
The practice of evaluating the marketing channels a consumer encountered on their path to purchase.
What would have happened if that specific channel was not there?
Multi-Channel Funnel
Shows how previous referrals and searches contributed to the sale.
However, it still not shows what would have happened if the channel was not used, or if a channel was different.
What is the core of a good attribution method?
Being able to answer the question:
“Would the customer had a conversion without this channel?”
Average attribution
Attribution method that assigns credits equally to all channels the customer came into contact with.
Time decay attribution
Attribution method that assigns credit across multiple events, assigning most credit to those closer to the time of conversion.
It is based on the hypothesis that those events have a greater impact on the decision to purchase.
First-click attribution
Attribution method that assigns 100% of the credit of a sale or conversion to the first channel that a user clicked through.
What is (also) not taken into account with last-click attribution?
- Banner impressions (without click, but view has influence as well)
- Non-observable channels (offline; tv ads, flyers, store visits).
What would the ideal attribution channel look like?
A channel that takes into account
- dynamics
- heterogeneity
- interactions/synergy
- alternative channels.
Limitations (disadvantages) last-click attribution
- No dynamics (between multiple sources of contact)
- No interactions or synergy (between channels)
- No heterogeneity (between customers)
- No alternative (no channel could have replaced the last one).
Last-click attribution
The channel the customer came into contact with right before the conversion is given full credit for this conversion.
It is the last advertisement a customer engaged with before purchasing.
It is used to see to which type of brand exposure a consumer responds.
Ways to make an estimation of what would have happened if the attribution method (advertising channel) was not there:
- First-click attribution
- Average attribution
- Last-click attribution
- Time decay attribution
Advantages individual level ads
- Many more observations.
- Ability to investigate how impact differs between customers (heterogeneity).
- Ability to fine-tune online advertising expenditures.
Individual level data
Disaggregated individual results. They are used to conduct analyses for estimation of subgroup differences.
Aggregate level data
High-level data acquired by combining individual-level data.
This data is about populations, groups, countries, or regions.
