This document describes building a custom attribution model to more accurately measure the impact of multi-channel marketing efforts. It outlines problems with traditional last-click attribution models and describes building a custom model that assigns credit to first sessions, last sessions, and all other sessions based on pageviews and time to purchase. Applying this model to an omnichannel retailer increased ROI of PPC advertising by 17% and identified 2.4 times more profitable keywords.
3. Why do We Need Attribution?
A typical customer purchase journey
4. What Attribution Model do You Use?
1. Position based (Last Click)
2. Probabilistic
3. Custom
4. Assisted conversions
5. Several attribution models at the same time
6. etc.
8. Problem
Marketing always means large business investments that influence brand
awareness and revenues.
10%
Of orders get lost
on the way to GA
50%
Of orders and more, are not considered when
evaluating advertising efficiency in multi-channel
companies
9. Problem
82%
Of orders are
made after 2+
visits
83%
Of sessions don’t
get evaluated when
using Last-Click
2+Devices are
used on a way to
a purchase
Position based attribution models don’t consider the influence of each
session on a customer’s way to a purchase.
10. Taking it to Another Level
1. How would online advertising ROI change if you consider offline
orders?
2. How does online advertising affect orders placed in a call center?
3. How much of the advertising budget was spent to acquire a
particular customer?
4. Which keywords drive the most profitable orders?
5. What keywords drive revenue in the brick-and-mortar stores?
6. What share of offline orders was influenced by online
advertising?
18. Attribution Logic
1. Determine the channel that introduced customer
to the brand
2. Determine the channel that converted customer
(Last Click)
3. First session gets 20% of the credit
4. Last session – 30%
5. Everything else – 50%
6. Consider pageviews during the session and
hours to purchase
20. Take Aways
1. Bring data together
2. Experiment – find patterns and connections
3. Apply ideas
4. Measure the impact
5. Rinse, repeat
21. Useful Links
1. How to evaluate ad success by and increase ROAS
by consolidating data
2. How to optimize ad campaigns with assisted conversions
3. Hoff success story
4. GCP free tier