Blog Post

Sailthru's Siloing

Sailthru's Siloing

In my last blog post Understanding Google’s Information Empire, I wrote about the technical side of Internet tracking, particularly the practices used by Google to drive its advertising network. Google is very possibly the most paradigmatic Internet company of all time, and it is often emulated, but not always. Google is a behemoth, but despite its relative stature, it is also an anomaly. Few other companies have the resources to offer high quality web services and also control all the advertisements that pay for them. Companies that would rather focus on their product than their advertising and marketing frequently employ the services of tech-centered companies to provide tools for tracking and communicating with their users. I met the CTO and a couple developers from one such company, Sailthru, at a hackathon recently, and though I was rather standoffish at first, they were all very nice and I’ve been communicating with them since. Sailthru’s product is a platform called SmartData, which makes it easier for both tech-savvy and non-technical marketers to collect and act based on detailed user data (of the kind discussed in my last post). However, Sailthru is uncommon in that it requires that its customers ‘silo’ that data on a per-site basis. While this means that Sailthru builds arguably less complete profiles of a user’s internet behavior, their policy gives them unique technical advantages, as well as the moral high ground. In an interview, their Chief Privacy Officer Caroline McCaffery noted that having company-specific information actually makes their technology more powerful – by keeping the information specific to particular brands, user info isn’t diluted - and companies can focus only on user interests and behaviors that are relevant to them. Here are a couple of the technical services that the company offers to marketers



Concierge– Boxes like these are beginning to pop up all over the Internet. When you reach the bottom of an article, links appear to other content on the website, thereby attempting to encourage users to continue browsing. They’re annoying, but also tempting.

Figure 1- Business Insider


Email list management– One of the company’s key products is their email-handling suite. They have templates and sophisticated software than can help companies differentiate their emails down to the level of individual users. Mass mailing and spamming tend to go hand in hand, but SailThru offers tools to ensure companies don’t send emails too frequently to particular users. Through SailThru, an email blast sent to a company’s whole email list might automatically leave off users who had recently received an interest-related promotion.

Figure 2-Sailthru Promotional


Tools for A/B testing – I didn’t mention this in the article about Google, but A/B testing is a common process of making minute changes to the web experience two sample of users have, and then attempting to evaluate the differences between how the two groups behaved as a result. Eventually, one of the two will be implemented.


Algorithms for interest – Sailthru attempts to provide quantitative measures of a user’s interest in a certain product or engagement with a specific brand algorithmically. For example, to judge interest on a news site like the one above, they might compare a user’s frequency of clicking links with specific metadata to the average across all links on the same page. If the number is twice as high, then the user is considered ‘interested.’ Similarly, if a user hasn’t logged in to a site in a long time after a gradual reduction in activity, Sailthru’s algorithm might flag that user as “disengaged” or “dormant”.

Figure 3 - Sailthru Promotional


As you can see, each of these services is directly dependent on Sailthru’s ability to collect user data. It’s hard to see how having more information about a user could really dilute the effectiveness of concierge, or the interest algorithms, but there are a few reasons that siloing data is actually a good business model, as well as a privacy-conscious one. In our conversation, Ms. McCaffery pointed out a business reason for siloing data – if information is your company’s stock and trade, it’s probably not a good idea to give it to your competitors. But the bottom line is that siloing data can make users feel more comfortable with being tracked on the Internet because it more closely resembles the implicit assumptions we make about privacy in real life. For example, let’s say I go to a restaurant and order clam chowder. If the next day, I decide to go back, and this time the waiter offers me clam chowder, I would feel happy that the waiter cares enough about my business to remember me. In that context, personalized service would be convenient and not creepy, because I consciously told him that information. I know the source and I know how it got from me to him. But let’s say the next day I decide to eat somewhere else – if I go to a different restaurant and a different waiter offers me clam chowder, I would feel slightly more uncomfortable. It would still be convenient, but I would feel anxious about how, why, and from whom the second restaurant had learned of my preference. For this reason, collecting personal data on other people’s customers can often backfire – for the companies sharing and receiving the data. A strict siloing policy like Sailthru’s allows for data collection on a human-scale, allowing web personalization without violating conventional notions of privacy. If such a policy were to be adopted on a large scale or at least encouraged by regulatory bodies, it could greatly improve the relationships between Internet users and companies that gather their data.


No comments