Dynamic Email Marketing: How Savings.com boosted CTR 88% with offers chosen by data, not instinct

by Adam T. Sutton, Senior Reporter

The coupon and deals website Savings.com uses an algorithm to select the offers it sends to subscribers each day. The dynamic content engine sifts through hundreds of thousands of deals to select 20 or fewer for each email.

Jacob Shin, Director of Online Marketing, Savings.com, helped launch the program. He says the company is striving for "one-to-one personalization," where each subscriber receives a unique email tailored to his or her preferences.

Five months after launch, the targeted emails increased average engagement metrics across the company's more than 2 million subscribers:

• Open Rate: 14% increase


• Clickthrough Rate: 88% increase


• Revenue Per-thousand Impressions (RPM): 147% increase

The company earns revenue through sales commissions and the advertising it displays on its website and emails. But not all the benefits of the new system come in clicks and dollars.

"From a qualitative stance, we're making that experience better for [users] so they come back to our site on their own and they remember the brand over our competitors,'" Shin says.

Below, we go through the steps Savings.com took to launch the content-choosing algorithm and how the team overcame challenges along the way.

Step #1. Start by personalizing the website

A challenge of running a good coupon site, Shin says, is to present offers that are:

1. Great deals


2. Relevant to the person's interests


3. Not expired

Savings.com created a system to automatically select deals that meet these criteria and prioritize them on its website. The system uses a series of calculations, called an algorithm, that combines several variables to gauge which offers are best for each member.

The algorithm considers:

• Expiration date of the offer


• Source of the offer (and that source's history on Savings.com)


• Relevance of the offer to a member's stated preferences


• Location of the offer in relation to a member's ZIP code


• And other variables

Offers that score higher in the algorithm are given prominence on the website.

The team has tweaked the system over time and maintains the ability to override it. For example, the team can force the system to prioritize offers that are:

• Related to an upcoming holiday


• Being promoted by a retail partner

Step #2. Send unique, custom emails

Savings.com sends emails to two types of subscribers: personalized and anonymous. Personalized subscribers have provided information to receive targeted offers. Anonymous subscribers have not selected preferences and receive general emails.

The company sends two different types of emails:

Email #1. Daily email

This message is sent once per day, Monday through Saturday, and it features deals with a shorter shelf life. Some of the deals expire in less than a day and are "hot and trending," Shin says.

Email #2. Weekly email

This message typically features deals with at least a 24-hour window of opportunity, Shin says.

Both of the emails include the person's username (if they provided one at sign-up).

Testing the new emails

The team added its offer-choosing system to its email program last year. Tweaks had to be made, but very little about the core calculation changed, Shin says. The team has learned the following works best in emails:

Fewer than 20 offers

A big difference between emails and webpages, Shin says, is that webpages can feature more information. The emails have to be trimmed to fewer than 20 offers.

"We have found from testing that anything beyond 20, 25, just becomes too long and it's easy to get lost in the email."

File size is also a consideration, he says. Keeping the number of offers below 20 helps prevent problems in deliverability and load time.

Diverse offers

Featuring too many of the same type of offer can put a cap on engagement, Shin says. His team tries to feature a broad selection of offers that are relevant to the subscriber.

"We found that the more diverse the content, the better the interaction from the user. So we try to include several different categories, whether it's a traditional online deal, a local deal or grocery deals."

Include some content

The team has also found that including helpful content can help lift overall engagement. In one recent example, the team provided tips on how to save money on Valentine's Day shopping.

Step #3. "Upsell" to free membership

The team has more than 2 million subscribers, and roughly half of them are anonymous. Since the content engine is shown to increase engagement, an important goal is to convert anonymous subscribers into full-fledged members. That way, the team can send better offers and boost results.

But "upselling" these subscribers into memberships is not simple, Shin says.

"We've found that some people want it to be quick and easy, and they don't want to invest a lot of time into doing our profile. Even if we say 'it takes less than 30 seconds,' we just always have a portion of our audience that's not personalized."

To encourage the conversion, the team features ads in its emails that encourage anonymous subscribers to become members. Once clicked, the ads take users to a four-step wizard tool to indicate their preferences. Once completed, the formerly anonymous subscribers are considered "personalized."

Give subscribers the option

If subscribers respond better to targeted offers, why not force them to select preferences?

That approach may earn conversions, but it could also cost the team many good subscribers. Shin's team avoids such aggressive tactics, he says.

"We want to give a good experience. … We have an almost 10% opt-in rate [for subscription], which is completely organic. And [the request to opt-in] is very subtle, it's not in your face."

Step #5. Make steady improvements

The content-selection engine is a major contributor to the success of the website and email marketing at Savings.com. Shin's team has the support from the CEO on down to make improvements, and almost all of the work to improve the system is done internally.

Here's how the team confronts two on-going challenges:

Relevance: often sought, never attained

Relevance is a theory, not a location. You can't "get there," but you can run toward it. The team works to improve its algorithm to make better offers, but it will never be perfect.

The team has added the following features to push the system forward:

• Voting - members can click "good deal" or "bad deal" to help rank offers on the site


• Change preferences - members can click "not interested" links in an email to show that they no longer want to receive emails about a particular brand, category or type of product

The team also plans to have the system consider the nature of an offer that a subscriber responded to. For example, someone who responds to an offer for a television will not likely need another television for at least several months. They may be better served by ads for other products.

Manage the volume

Selecting offers for each subscriber is a demanding process that takes several hours for the team's servers to complete each evening. The scope of the system presents a unique set of challenges.

Shin's team has an engineer dedicated to ensuring the system's calculations are accurate, and that the results are sent without issue to the team's email service provider (which compiles the emails).

Useful links related to this article

CREATIVE SAMPLES:

1. Daily email


2. Weekly email


3. Preference Selection Wizard Tool

Copywriting Contest: Write the best-performing subject line and win

Email Newsletters: Social media integration yields 135% more traffic for New York Public Library

Email Design: 3 critical factors of ‘lucky’ campaigns

Savings.com