A consumer goods company operates two brands that sell the same products to different target audiences. The company works from a database that comprises 15 million+ entries for customers/leads/etc. Each brand sends one email a day, mails 6 catalogs throughout the year on different schedules based on seasonal buying patterns, runs search/display/etc ads across Google and Bing every day, runs Facebook/Insta/Messenger/etc ads every day, posts on various social networks every day, works with hundreds of affiliate advertisers that advertise every day, puts time/effort/money into SEO every day, and just started running TikTok campaigns because the founder’s daughter’s friend is an “influencer” on the network.

Some people buy. Some of those people have bought before when they lived in a different state before they were married. Some people see an ad on their phone and purchase by telling their spouse to order from Amazon. Some people wait until their paycheck clears the bank two weeks later then buy, but not before they’ve seen the brand an additional 15 times on various channels.

Multichannel attribution aims to attribute those purchases to the correct channels, usually to determine ROI and help with planning.

Keeping track of all those customers and touch points is non-trivial. Assigning weight to each touch point is non-trivial. Modeling behavior accurately is non-trivial. It’s theoretically 100% possible. Software companies and consultants sell solutions that seem plausible. In practice it feels impossible.

Source: A decade of agency and in-house advertising/marketing work for some of the largest brands in the world