What Will They Buy Next? Using Product Recommendations in Email

Greg Zakowicz, Senior Commerce Marketing Analyst

Greg Zakowicz, Senior Commerce Marketing Analyst

Author Bio

Greg Zakowicz is a senior commerce marketing analyst at Oracle's Bronto Software. With more than 10 years of experience in email, mobile and social media marketing, Zakowicz knows the retail industry and its challenges, staying on top of the latest trends by leveraging deep insight into the marketing spectrum. His subject matter expertise stems from his experience in providing commerce marketers — including numerous Internet Retailer Top 1000 clients — with in-depth analysis of their marketing programs, recommendations for improvement, best practice support and implementation guidance and execution.

Zakowicz is a frequent webinar speaker and presenter at ecommerce events, such as Fashion Digital New York, SIA Snow Show and ROI Revolution Summit. He has been published by top retail and marketing publications, including Power Retail and Inside Retail, and is a regular contributor to Bronto’s Commerce Marketing blog. You can follow him on Twitter at @WhatsGregDoing.

Creating relevant emails is not a new concept, but due to resource limitations, it can often be difficult to execute. If you can’t be as targeted and relevant as you’d like at the individual email level, using product recommendations in your messages can fill the gap. For those who can create targeted promotional emails, product recommendations can help you take those messages to the next level.

When using product recommendations, however, there are things to be mindful of, such as what your recommendations are based on. Is it purchase behavior, click behavior, website browse behavior, or a combination of these? It’s important to determine up front so you can audit the results for accuracy and identify opportunities to potentially collect additional subscriber info. Doing so will help create more accurate recommendations.

Let’s look at some messages where recommendations would be most effective and some pitfalls to be wary of when doing so.

Lifecycle Messages

Inserting product recommendations in lifecycle messages, such as a post-purchase, anniversary or birthday messages, is a natural fit. These messages are already relevant by nature, and including specific suggestions based on customer history can make them even more powerful.

If you’re using a lifecycle message where you don’t have purchase info but may have click and preference details, such as a welcome series message, use whatever data you have to be as relevant as possible. However, be careful not to refer to your recommendations as being “just for you.” Instead, consider using something more diplomatic, such as “items we think you may like” or “customer favorites.” You’re still learning what they may be interested in, so don’t make it sound as though you’re already certain of their preferences.. Doing so might give a negative impression.

Cart Abandonment Messages

As consumers get closer to making a purchase, recommendations may be the deciding factor in not only getting the contact to buy but also increase their cart total. Maybe the contact sees a dress in the recommendations that they missed while browsing your site. Or it could it be that pair of shoes that complements the outfit in the cart. Don’t underestimate the power of recommendations, even at a point so close to purchase.

The examples below show three slightly different approaches to recommending products. Pottery Barn Kids showed “also viewed” products, while NewEgg labeled their section “Trending Now.” Do I really believe that all four things trending on NewEgg are TV-related items? No, I don’t. And Adore Me may actually be showing their most popular sets, but they may not be as relevant to the shopper with this message because they tried to cross-sell their primary products rather than recommending products similar to the abandoned pajamas.

Email ExamplesAll three examples were well-constructed, but I would avoid using labels along the lines of “Products Recommended for You” in an abandoned cart message. The subscriber already identified specific items they’re interested in, so your goal should be to either complement the abandoned product or provide a similar solution to the problem the abandoned product solves.

Transactional Messages

Don’t forget about your transactional messages, such as order and shipping confirmations. These messages can be substantial revenue drivers, and considering recipients have just made a purchase, it might be the perfect opportunity to showcase some upsell or cross-sell items here. In this Williams-Sonoma example, the “also purchased” products are mostly relevant. Why wouldn’t I need a new spatula along with my new griddle? What I like about this is they are also relatively low-cost items, so making an impulse purchase decision is more likely.

Williams-SonomaIf you can easily edit the message layout, I prefer to have this recommended content along the right rail as opposed to below the primary content. This allows you to keep your promotional content above the fold while remaining CAN-SPAM compliant.

Be Wary of Automated Promotional Messages

If you plan to send recurring recommendation emails to your audience, don’t automate the exact same message. Change the subject line from send to send. You could still automate your message, but be sure to keep your message fresh. All it takes is one not so relevant email before a contact decides to never open your messages again. And using the same subject line allows the subscriber to identify this message in the inbox before even opening it.

For instance, I once received an email from a company with the subject line “Gregory, Recommendations Just for You.”  Upon opening, not a single product was indeed relevant to me. Now, every time this company sends me that message with the identical subject line, I immediately hit delete. Sending a message with an identical layout is fine. Just be sure to freshen the subject line with each repeated send to a contact.

In this example from KarmaLoop, I am a non-purchaser. They continue to send me recommendation messages like this. As you can see, these products don’t have much in common, yet the creative says these are just for me! The products are for both genders and from a variety of price points. This may be an attempt to determine what I click on to further their segmentation, but they could do that with standard promotional messages. They could also collect gender at sign-up, and even with no other info, make these recommendations more relevant.

Subject Line: Just for you. Really. (this week)

KarmaLoopWith Overstock.com, I have purchased in the past. I once bought a couple of iPhone accessories for my wife. What I don’t necessarily like about this message is the primary recommendation for stylus pens with a sub-recommendation for an iPhone accessory. Based on my purchase history, the iPhone accessory is more targeted to my potential needs, and the other two jewelry recommendations miss the mark. I have no issue with trying to determine whether I might be in the market for some jewelry, but two recommendations for jewelry seems like a stretch to me. Why not include one jewelry item and maybe one other female fashion accessory?

Subject Line: Gregory, recommendations Just for You…

Overstock exampleI do like the price points of the suggestions. They are all in line with the price of the original items purchased, so they are not making drastic recommendations. But the subject line used for this message is identical to the one I mentioned from another retailer above, so I know this is an automated message. Again, if these recommendations completely missed the mark, I would be very hesitant to open in the future. In this case, it was relevant enough to open again.

Always check that your recommendations are indeed hitting the mark. Remember, results will vary based on what you  segment on (browse, purchase or click behavior). However, if they don’t appear to be delivering the results you’d expect, edit your settings and keep tweaking until you get the best possible results.

Final Thoughts

People have different needs and shopping behaviors, so recommendations don’t always have to be exact. They should simply estimate what people may like. But with that being said, they should still be in the ballpark in order to engage the subscriber. Test and monitor the results of your recommendations for relevance and accuracy. And finally, don’t force these recommendations into all of your messages. Look for the opportunities where it makes sense to do so and gives you the greatest likelihood of success.