The problem is that Google AdWords Conversion Metrics tell you the number and value of sales that come as the next action after landing on an Ad. What they do not tell you is how significant that action was in the stream of events that preceded each sale, and our finding is that the average number of events prior to a retail sale is between three and four.
Another issue here is that Google only tracks interactions with digital, so no account is taken of retail visits, call centre logs, catalogues sent, email opens etc.
This results in Google reporting a very inflated view of the significance of their ads.
One retail client of ours was spending a considerable proportion of their marking budget on Google Ads, but did not have confidence in the sales results they were reporting.
We analysed their Google Analytics data feed and discovered that by looking at just the online events, and just the last clicks, as Google does, that Google Ads (or Pay per Click as it used to be) could claim an astonishing 39% of the value of all sales.
However, when we took the Google Analytics event feed, and merged it with all the clients other online and offline events data, the UniFida approach to multi-channel attribution got to an entirely different attribution outcome. In this case the Google PPC share of order value was reduced to 8%.
The UniFida approach is to look at all the events that take place in a 90-day window before a sale and then to use AI to weight the significance of these events. In this way something that happened through a specific channel 30 days before a sale gets a weighting that is different to something that happens via another channel on the day of the sale. This weighting is determined by statistical analysis using artificial intelligence developed by our data science engineers.
Because customers control their own journeys, reliance on first or last click as the only driver behind an order is absurd, hence the requirement for AI to attribute the correct significance to each event. Almost half of all advertising expenditure is offline so, omnichannel attribution must include offline as well as online.
Are you interested to find out which parts of your marketing are not working as well as the bits that are?