Consumer behavior is fundamentally different at this time of year, which means that marketers may not be able to use the same data and algorithms that they used in August or even October.
Lack of preparation and adjustment can be very costly during the holidays unless you make three changes.
- Update your algorithms to accommodate changes in behavior
I say this over and over: AI is really just math, sometimes complex math, and it’s up to analysts to use the right data to make it count, especially during the holiday season. Machine learning algorithms are not healthy; without hesitation, use and believe in the data provided.
The customer’s behavior is very different during the holidays: you receive an influx of buyers who are not involved at a different time of the year; customers look for specific brands and categories that they are not yet familiar with; they will buy faster than normal, and the list goes on. Also, everything about their behavior is now expanded: the number of times they visit and leave your site, the number of products and categories they are looking for, the number of items they want to list and add to the cart, and so on.
- Find out how your marketing campaigns will be affected
Since your machine learning algorithms don’t apply to holidays all year round, you’ll also need to adjust your campaigns.
For example, your most likely-to-buy campaigns will look very different, because just having someone on your site for a significant amount during Black Friday and Cyber Weekend does not mean that you are unlikely to buy. While this type of interaction traditionally leads your algorithm to think that visitors are likely to buy more and then suppress certain promotional offers, you want to do the opposite during the holiday season and show those offers when those potential customers demonstrate more engagement.
Your uninteresting customer campaigns are also affected. Traditionally, a customer spends several hours or days with your brand to determine what to buy and whether to buy. During the holidays, this period is reduced to a day or two – they want to make quick decisions and look for prices and comparisons. During the holiday season, performance should be achieved to disconnect customers, when they are on-site for hours, not days.
- Prepare data for the end of the season and other holidays
Keep the data set separate after the holidays. You do not want to insert the specific holiday into your regular campaigns; This will affect your specific Christmas clothing patterns.
For example, your normal and loyal purchases will likely go to your website to make purchases for other people, so what they look for and buy during the holiday season can be completely different from what they usually see the rest of the year.
That is why you define holidays as a completely separate period of behavior and then define your patterns to reflect the most typical patterns of behavior you see from your customers throughout the year.
However, you must update and use these holiday templates for other important purchases and gift holidays. Valentine’s Day, Mother’s Day, and Father’s Day, back to school, and especially Cyber Week 2020 are times when you should consider creating and implementing specialized models based on what you have learned during that time.
In general, end-of-year shoppers are enthusiastic and short on time, forcing them to make faster decisions, give up their typical shopping and browsing habits, and make more expensive purchases than at any other time of the year. Your seasonal algorithms and campaigns need to adapt to these changes, and you still need to focus on recent (versus historical) engagement.
If you haven’t already, make sure your data science and marketing teams are aligned and update your machine learning strategy accordingly.