Skip to content
Roost's Tim Varner on how push notifications can help online stores

Roost's Tim Varner on how push notifications can help online stores

Whatever the experts say, email isn't dead. However, it is true that inboxes are getting more and more crowded each year. This has led marketers to try and find new ways of reaching people—and one of the marketing channels on the rise is push notifications.

Roost is a fairly new tool that lets site owners engage visitors with web push notifications. If you use Chrome or Safari on desktop, you’ve probably seen an alert beneath your search bar asking if you’d like to allow notifications from a particular site. For visitors who opt in, Roost can send on- and off-page notifications based on a variety of triggers.

Co-founder Tim Varner says Roost grew out of an idea he had while working with an ESPN content network. He saw mobile browsing rising from month to month, with apps becoming the defacto solution for mobile engagement.

“Those were all small sites,” Varner says. “They didn’t want and couldn’t afford a standalone app for 36 different content sites.”

At the time, Varner was contributing to SXSWorld, the quarterly magazine of South by Southwest. While at the conference in Austin, he met Burton Miller and Casey Haakenson, who were working in native push but “daydreaming about the possibility of doing push for the web, because they saw the web as a much more needy market.”

“They wanted to build something to fill that technological need, and I saw a need from the perspective of publishers,” Varner says. The three became friends and developed their idea, raising funds and launching Roost in January 2014. “It was really just a meeting of minds between three people who saw a great need for push on the web.”

I caught up with Varner to ask about how Roost works, what kinds of specific benefits it has for ecommerce merchants, and what he sees in store for ecommerce in 2016.



Push notifications are getting a lot of buzz as a new marketing channel. How do you see Roost being used in ecommerce environments?

There are basically two use cases we see in ecommerce. The first is a more general use case—whatever content strategy the site has, to be able to broadcast that content via notifications, basically like a content distribution channel, similar to a publisher’s.

The second use case, which we sort of split into multiple use cases, is to do automated notifications based on inventory or pricing. That’s easily accomplished with our API. And then kind of embedded in there is deep segmentation, based on browsing behaviour. It costs more, but we can segment your audience based on purchased items or inventory viewed. Notifications work best when the target audience is more carefully segmented.

We provide notifications 1:1 at scale—personalized messaging, in other words. It would be something like, "[Name], come check out the [item] you looked at."

With ecommerce, we’re interested in things like cart recovery and pre-scripted notifications that go out at preset time intervals, like "Tim, come back and finish your purchase" or "Hey, did you forget about [blank]?" or whatever. Maybe a brand  will attach discounts to the notifications. That's up to the brand.


Are there examples of segments besides purchases and browsing history?

We have a system called aliases. If you have any unique identifiers for a customer on your end, you can pass us those identities, and they can become a segment as well.


Is Roost capable of running more triggered, evergreen-style campaigns?

Absolutely, yes. Imagine anything you can do with a basic email marketing suite, from a campaign standpoint—you can do the same with notifications.


On a user-by-user trigger basis, as well as on a global scale?

You can do either/or. It could be a thousand people in a segment, or it could be one person in a segment. That totally depends on how you want to use Roost.


How do you see personalization helping ecommerce merchants?

Audience re-engagement and more personal messaging. Personalization is essentially a stand-in for optimized relevancy—it's a much more relevant notification. It's addressed to the person in many cases, and then it contains content that is particular to that person's tastes or preferences.


What makes offsite notifications that much more advantageous than email?

Getting to someone's lock screen or to their monitor when they're offline—these notifications have a much higher click-through rate. It's about 23 percent, which is much, much better than email. Even a great email campaign won't get nearly that click-through rate.

Push notifications give you the ability to reach someone when they're not online and create a new session. On mobile, it's almost a 1-in-4 likelihood that you will create a new session, which is pretty great. The click-through rate on desktop is also far superior.


Do you see challengers for this marketing space? Do you see it filling up eventually?

We think web push has a very big future, but we see it as a channel that you would use in concert with many other channels. In fact, it's probably best used in coordination with (rather than in contradistinction to) other marketing channels.

For example, you could attach a UTM to the link you put in an email campaign, and then we would watch that and say, “If that link is not clicked within 60 minutes of email receipt, we’ll send a notification at minute 61 with the same offer driving traffic to the same landing page, and then we can do pre-scripted notifications coordinated with that email campaign. And so, in that sense, you're not waiting until the next day or two to be like, "Oh that email campaign wasn't very effective"—you know it within a certain time period, and you can start a drip campaign on push to help capture some of the audience you didn't gain from the initial email.

Another example is our A/B testing feature, which a lot of brands use as a social leader. They give us, say, five headlines, and we send out that notification. Within a few minutes we're able to get the winner and then they can craft their Twitter and Facebook messages based on that, because the length of the push and social messages are roughly approximate. It gives you a really good sense of what messaging is going to win an audience.

We really like to think of Roost in terms of how it can be used as part of a more comprehensive marketing campaign, not as a way to do away with other marketing.


What kind of results have you seen for your Shopify customers? What might Roost look like for a Shopify merchant, for example?

We have a Shopify plugin. We had a customer who asked for it, and we built it for them, and they've been using it. They've had a successful implementation. But at the time it was really a test case.

The question we're really interested in, within ecommerce specifically, is when is the best time to trigger the opt-in? When the customer is on a product page? When they've put something into the cart? When they've just finished an order? The moment they land on the site? You could have user-generated opt-ins at any number of points. And then, from a campaign standpoint, what are the best baseline campaigns to set up right out of the gate? Is it the cart recovery campaign?

Is it a more general, all-purpose seasonal campaign? Those are the sorts of things we would love to uncover. Right now we don’t have enough ecommerce sites using Roost to get those numbers. From a global standpoint, things look really good. The global opt-in rate on mobile is about 6 percent. On desktop it’s about 4.5 percent, which is huge when you consider that that's against your entire web audience. Ecommerce companies go through great pains to get email addresses—they certainly don't get five percent of their traffic giving them their email address. And then click-through rate—on desktop, it's in the single digits, about 8 percent—and on mobile it's 22 percent, a huge number.

We're really happy with those numbers. We actually think—this is a bullish bet—we think that the numbers would be much higher in ecommerce than they are with publishers, because in the case of publishing it’s possible for the end user to receive the notification and get 100 percent of what they want right there within the notification. On the other hand, we have gaming sites using Roost, and in some of those cases the click-through rate is almost 40 percent, because the user has to actually go to the site to interact with the content they want. One of the reasons we want to push into ecommerce is because we’ve seen how well sites that require an onsite interaction perform versus sites that don’t.


Could a Shopify merchant use Roost to provide a personalized customer service experience? For example, a couple weeks after a customer buys your product, you could send them a follow-up notification.

I actually had not thought about that. Just off the top of my head, that seems like a totally viable use case. There would be no technological barriers to putting together a campaign of triggers based on that. That is totally a viable use case, one I hadn’t thought of before. Pretty good.


Do you have any idea when iPhones will be able to offer this functionality?

We have a really good idea on everything but iPhones. We do onsite messaging—that is, we can send notifications if the person has the ecommerce shop open in a tab somewhere and you're online. We can support every browser on every device right now. For offsite messages, we have Chrome on desktop and mobile, Safari on desktop, and we're about to add Firefox on mobile and desktop.

iOS is more of a mystery. We have some reason to believe there will be a mobile spec for iOS in 2016, but that bet is speculative.


Where do you see ecommerce going in 2016?

Interestingly, I think one of the under-told stories of 2016 will be the web going to HTTPS. Not that the entire web will get there in 2016, but by the end of the year I think we're going to see Google—Chrome in particular—throwing security warnings for sites that have mixed content. That actually includes a lot of ecommerce sites. It certainly impacts all of publishing. How are sites going to deal with mixed content issues? How are they going to deal with the need to be 100% totally secure and positioned to prevent man-in-the-middle attacks?

I'm interested to see what ecommerce companies and brands do there. Each vertical will have a different path to satisfying those requirements, which could impact engagement, as well as audience gain and loss. For example, publishers that don’t switch over are going to start losing audience when a significant portion of their visitors get a security warning when they land on their site. For ecommerce it would be the same thing, with customers getting a big scary "This site may not be secure" warning when they get to the checkout page.

Another thing I’ll mention is the mobile web browsing audience, which I think has been sort of under-messaged. It's not an audience that has had a lot of re-engagement tools. So it's kind of interesting to us, to what extent can web push engage an audience that's been sitting somewhat dormant. In publishing and ecommerce, most sites get their referral traffic on mobile from deep links within Twitter and Facebook’s apps.

What we like about what web push provides is a 1:1 messaging channel between the brand and its audience, and you don't have to sell the audience or rent it out to a third-party platform—you can engage them directly with your own site. Look at how much Buzzfeed spent on Facebook advertising at the beginning of 2014. Something like $6 million for the first half of that year? It seems unsustainable to me for brands to spend that much on Facebook and Twitter advertising when they can have that same engagement directly from their website.

To learn more about Roost, visit

Previous article Superstore vs. Empire