Net Conversion is Not Conventional. Part of #staycurious means challenging the marketing status quo and finding that sometimes, the norm isn’t truth. Here are three marketing and analytics myths that we busted this year.
Using consumer logged-in data and household wi-fi IP-matching, bottom-up attribution can allow integrated media strategies to reach consumers on all devices.
When the same user is signed into Google or Facebook on both their phone and laptop, they can be tracked cross-device as well as via GPS.
When logged into the same WiFi, a user or household’s online actions can be tracked from TV ad view to website visit.
We work with resort clients that don’t use an online booking engine for reservations, and instead rely on call agents for revenue tracking and recording booking data. This might stump another agency, very limited conversion signals toward which to optimize their digital campaigns.
Not for us. We set up a script in Google Analytics to programmatically pull user information to determine what behavior correlates. We compared offline bookings for different months with event actions, event categories, a combination of both, and pagepaths.
We identified some unexpected behaviors that had high correlations with bookings, including user interaction with the resort map online, and visits to a select welcome page that website users could not navigate to organically. The welcome page was emailed to consumers after making a booking. We optimized our digital campaigns toward this page by including it as a conversion signal, and have since been able to report even more accurately on a more holistic snapshot of the consumer path to purchase. Even when that path is in the dark.
How can they be when what you’re trying to do is predict the future? Turns out we can get pretty damn close. Within 0.1%-0.5%, actually.
We used over three years of historical data for one of our clients paired with multiple external factors to create a tool to generate extremely accurate revenue forecasts beyond using simple year over year performance.
We created a daily automated forecast model relying on Python and automated data feeds from Adobe Analytics, Data Studio, Big Query, and our own internal software Conversionomics. Using factors such as public holidays, school holidays, actual sales for historic arrival date, cumulative revenue, seasonality, recent trends, and prior growth, we can revenue up to six weeks in advance.
Because our model forecasts out a significant period of time, with significant accuracy, we allow our client time to plan for soft periods. We’re also able to use the model as a media planning tool to reallocate marketing budget, as it lets us identify markets that show need for upcoming periods.
Additionally, beyond a more accurate revenue forecast, we’re able to compare the accuracy of previous forecasts against current state to continually monitor and improve the model’s success. Our forecast is so tight it’s leveled up to a fivecast.