What personalized data delivers the biggest ROI for retailers?
Through a special arrangement, presented here for discussion is a summary of a current article from MarketingCharts, a Watershed Publishing publication providing up-to-the-minute data and research to marketers.
While personal data (name, gender, location, etc.) is — unsurprisingly — the data type most commonly used to personalize the web experience, personalization based on purchase history has the biggest ROI impact, according to a recent survey from Econsultancy.
The findings were based on a survey of more than 700 marketers and agencies, roughly half of whom are based in the UK.
Asked to rate the ROI impact of personalization based on the various sources of data, almost three-quarters of company respondents said that personalizing based on purchase history has a high impact on ROI. Given that only around four in 10 respondents are currently personalizing based on this data, it appears that this is the area with the most opportunity.
Meanwhile, personalization based on user preferences (70 percent) and web property behavior (68 percent) also appear to have a high impact on ROI, while most see mobile app behavior and third-party demographic data having low or no impact.
Turning to the channels through which marketers and agencies are personalizing the digital experience, the report notes that e-mail (78 percent marketers; 80 percent agencies) is the most popular, followed by desktop websites (69 percent and 78 percent) and mobile websites (36 percent and 51 percent). Fewer than one-quarter are personalizing the paid search experience, despite a separate Econsultancy report finding that personalization’s impact on conversion rates is highest for search engine marketing.
Sticking with conversion rates, the study finds that they are the leading metric for analyzing the benefits of personalization, used by 65 percent of company respondents. Customer acquisition is the second-most used metrics, ahead of customer retention.
Discussion Questions
Do you see any other data sources coming close to purchase history in helping retailers personalize the web experience for shoppers? What has impressed or disappointed you in how websites personalize your web shopping experience?
A big one might be consumer or shopper insights from brand supplier partners, be it their aggregate analysis around intent or segmentation based on retailer data, or the first-party profiles developed through one-to-one digital marketing efforts.
Maybe we need a new word. “Personalize” from “personal” means “aimed at some particular person (usually in a hostile manner) first attested 1610s”—at least according to the online etymological dictionary. The part in parentheses the key.
Am I the only one who doesn’t value this “personalization?” Does no one else feel used, violated and deceived? The whole privacy movement is gaining momentum, thank goodness. You can see this in the backlash against Uber, for example, for their abuse of personal data. I’m in a Yahoo group of Mac users who use this vehicle to vent their anger with Apple. Someone in the group asked about a particular scanner the other day and I’ve now received three ads from Amazon about that very same scanner.
Please retail, stop thinking you know me. And stop telling me what I should buy. You may think it’s giving you ROI, but like always you’ll overdo it and it will come back to haunt you. I’m waiting for an email that begins with “We noticed you didn’t finish your broccoli at dinner last night … “
There are issues beyond the data that go back to the merchandising decision. If I know purchase behavior, I still might not know the right promotional instant message to send you. If I know that your weekly shopping basket includes Coke, do I send you a coupon for Pepsi to try to get you to switch, send you a coupon for Coke and make you feel special (but erode margin), or try to get you to buy a complimentary product, which may or may not seem to be an irrelevant offer and just end up bugging you? So having the data (if you can get enough of it to have a complete picture) is one thing, figuring out the next best offer is yet another.
Personalization is most impactful when it creates relevance in the customer’s brand experience. Beyond purchase data, other behavioral data is also helpful. This includes service or campaign data (retargeting for example) as well as browsing history, data gleaned from a customer’s profile, preferences or social graph.
The majority of e-commerce sites still do a poor job of leveraging all these data. It’s ironic as this is an area of opportunity to not only drive sales but also build the customer relationship.
For a few years now, there have been technologies available to uncover the insights we’re talking about here. Consumer shopping “missions” continue to evolve. Consumers often start by looking for items in a single product group, but end up buying items in other product groups as well. Sometimes they purchase such additional items in specific patterns. The sources for finding this information are multiple. The channels to get to these sources are also several in number. I think there are enough IT vendors in the marketplace today to help merchants find this information in a productive, affordable manner.
In terms of my own experience with merchant websites, as long as I don’t get bombarded with irrelevant ads, I do appreciate personalization capabilities to make my shopping experience more satisfying.
Personalizing based on purchase history reminds me of sage advice known to any commodity trader: “Past performance does not guarantee future results.” Routine purchases of branded products likely illuminate a pattern, but to predict is something else. I find it aggravating when Amazon places ads for items I just looked at, or when they associate a recent purchase with another suggestion, when the relevance or context of that purchase are unknown to Amazon. Impressive technology, yes. Relevant, not often. How about using data to introduce new ideas, or new ways to use a product I’ve used in the past? Personalization with anticipatory relevancy I could get into.
There are two approaches. We can infer information from past purchases, searches and additional contextual clues like location and touch point. Or we can ask a series of questions to build a detailed profile. Inferences tend to get polluted with gift purchases, multiple people in the same household sharing an account, etc. Straightforward asks would seem to be the most accurate.
If I bought a faux fur vest last week, it does not mean that I want another faux fur vest this week or any other week. Same for the WiFi extender I bought yesterday. Quit showing me the ads, I bought mine.
If retail is going to look at purchase data, it would be far more helpful for them to determine that I like accessories including vests and scarves. They might also be able to categorize the type of items I tend to pay full price for as opposed to shopping the sale racks.
Let’s apply a little more brains to the process, how bout it? For my 2 cents!
Shoppers want retailers to make shopping easier and more personalized. Using shopping history history to offer new items that are complementary to previously purchased items. Based on previous clothing purchases, for example, the shoppers size, style preferences, color preferences, accessory choices, etc. allow the retailer to offer similar ensembles or accessories for previously purchased items. Or, in the case of books/music, preferences of genre, author, etc. (as Amazon does today).
Allowing shoppers to indicate when they are purchasing for themselves v. someone else helps to tailor more precisely (Amazon offers me titles that are of no interest to me because they can’t tell the difference between items I buy for myself v. gifts). Keeping a digital collection that the shopper can reference online or in-store makes it easier to find items that work together and are compatible. Using the customer’s own data to help the customer—that’s personalization that is useful, not creepy, and increases sales.
As the chart shows, personal data breaks out into (1) who customers are, (2) what customers say, (3) what customers do.
Without question, “what customers do” behavioral signals have the greatest ROI. The question is around which behavior is most predictive.
Grocery answered the question long ago. Purchase history is wildly predictive. This is largely due to short purchase cycles and repeat store visits. The same can be said of entertainment, with the emergence of streaming video.
Products with longer purchase cycles, such as autos, can’t integrate purchase history as easily. They rely on other forms of behavior, such as searches, site paths, and inquiries. These are powerful signals of intent.
Also in this category is real estate, where “what customers say” is critical for agents seeking homes.
The problem is getting enough of this data. Large segments are completely uncharted. Faced with vast blind spots, data on “who customers are” becomes all the more critical for long purchase cycles like like auto and home.