Case Studies: Intermix

The Company: Intermix, women's apparel sold online and through the company's retail stores

Since its inception in 1993, Intermix has become synonymous with innovative dressing and highly evolved personal style. The concept is simple, a "mix" of established and emerging designers at a range of prices, displayed in unexpected combinations, is what sets Intermix apart from other retailers. Intermix currently has twenty four store locations throughout the US, with the newest locations in Los Angeles, Chicago, Charlotte, Orlando and East Hampton.

The Problem - Cost effectively continue and increase revenue growth even in a down retail market

Don McNichol joined Intermix in 2005 as Director, E-commerce to spearhead the company's multi-channel expansion drive through the launch of IntermixOnline.com. McNichol needed to find ways to enable Intermix to cost-effectively continue revenue growth while increasing the company’s profit margin.

McNichol says there are traditionally 3 key levers in the control of a retailer: traffic, relevancy and margins. Understandably, McNichol and Intermix wanted to find a way to cost effectively increase all three.

McNichol says there are traditionally 3 key levers in the control of a retailer: traffic, relevancy and margins. Understandably, McNichol and Intermix wanted to find a way to cost effectively increase all three.

Before adopting Autotarget, Intermix would simply send email blasts to everyone on their target list. A common scenario involved a customer receiving an email on Friday, looking online after work at the specials and new products, discussing what she had found with friends on Friday night and shopping on the weekend.

While email marketing was profitable, McNichol felt there was a big upside to more effectively targeting his email campaigns. He, especially, wanted to find ways to email more often to customers who wanted frequent contact from Intermix while both minimizing discounts and increasing the lag time of discounts (thereby shortening the discount period).

In order to achieve Intermix’s revenue and profit goals, McNichol, therefore, needed a practical way to identify who to give a discount to, how much of a discount was necessary, how long Intermix could wait before offering a discount and how often he could email specific segments of customers.

The Solution — Autotarget for segmentation and predictive analytics

McNichol turned mid-2008 to a new product, Autotarget, developed by his email service provider, iPost. McNichol felt that Autotarget’s Recency, Frequency, and Monetary value (RFM) segmentation capability could be the answer to his problem. Peter was familiar with RFM from its decades-long track record of successful use in catalog and other direct mail marketing, where these three criteria have been the most reliable predictors of future customer engagement.

Autotarget drew behavioral data from two Intermix sources: email responses from the iPost email tracking database and Intermix purchase behavior data. Autotarget's RFM analysis automatically placed Intermix's customers into micro-segments (called cells) based on past behavior, including click rate, open rate and purchases.

McNichol used the Autotarget predictive analytics data to send a targeted test mailing to customers in specific cells and was able to easily segment customers into three groups:

  • VIPs who represented 20% of the customer base. These customers have a higher amount of disposable income, will shop for latest trend and demonstrate low price sensitivity.
  • Sale Shoppers who represent 40% of the customer base. These customers are "aspirational". They like great garments but, also, demonstrate high price sensitivity.
  • Brand Shoppers who represent 40% of the customer base. Brand Shoppers shop specific designers, like a good deal but only demonstrate medium price sensitivity.

McNichol started conducing A/B tests to determine the most effective offer for each segment. He identified three approaches:

  • VIPs - usually full price. Occasionally Intermix will offer something like a $200 gift card instead of a discount but primary Intermix offers VIPs specials such as private events - e.g., be the first to come in, get a glass of wine, check out a new line and meet the designers.
  • Sale Shoppers - 30% discount. Now Intermix only needs to offer steep discounts to this segment of the base.
  • Brand Shoppers - 10 to 15% discount. The most effective approach is a combination of information on specific brands and special discounts to spur action.

The Results - More revenue, greater profit and greater understanding of customers

  • Profit: Email marketing profit increased +28%
  • Revenue:
    • Autotarget has increased revenue by 9% (note: Intermix started using Autotarget mid-2008)
  • Discounts: 40% lower on average for a 3 week shorter period of time
    • Before Autotarget (2007 - mid 2008): 30%
    • After Autotarget: 18% on average
    • Change: -40% (decrease) with 3 week greater lag before sending the discount
  • Opens:
    • Before Autotarget (AT) (2007 - mid 2008): 20%
    • After Autotarget: 38%
    • Change: +90% (increase)
  • Click Thrus
    • Before Autotarget: 35%
    • After Autotarget: 51%
    • Change: 46% (decrease)
  • Opt Out Rate:
    • Before Autotarget: 1.2%
    • After Autotarget: .19%
    • Change: -84% (decrease)