Monthly Archives: March 2011

by adgenie

Search Retargeting & Magnetic Part 1: The Two Jobs of a Great Engineering Team

It isn’t hard to explain at a high level what Magnetic does. In the simplest formulation, we use search data to target online advertising across the Internet. That is the mission of our business.

To succeed in our mission we need to solve many engineering and business problems. First, we must provide our customers with access to enough of the right data for them to effectively target to meet their campaign goals. To do this, we give marketers a scalable, powerful and yet simplified solution to create, manage and review the success of their campaigns. Additionally, with billions of targeting actions per month being executed across Magnetic’s platform, we need to analyze our data in order to identify ways to better serve our customers.

So, while the high-level description of our business might seem straightforward, the engineering and product problems aren’t so simple. Above all, there are two rules we use to distill the responsibilities of Magnetic Engineering into a core team mission statement: keep our products running, and continuously improve our products.

DevOps

Keeping things running has traditionally been in the realm of operations, the engineering sub team responsible for keeping systems available and performing well. At Magnetic we follow the principles of DevOps, which requires developers to be responsible for supporting the software they create, and in general, integrates operations with development to a great degree. As we grow, there will be dedicated operations engineers, but application developers will never stop supporting operations.

All engineers at Magnetic are responsible for supporting applications released to customers. Developers test releases for performance and performance under load. Additionally, they use automated tools to release their software to customers and build monitoring and alerting into their applications

Product Development

Product development serves as the engineering team function focused on improving products and developing new ones. Magnetic uses the Scrum methodology for product development. Agile approaches such as Scrum posit that the engineering team’s primary responsibility is to regularly release software changes that increase the value of the company’s products.

Agile is a great fit for Magnetic. Most importantly, it keeps the team focused on what matters most — our customers. By releasing often, we quickly deliver new features, smoothly integrate with new business partners, and proactively make infrastructure improvements that keep our systems working well.

Agile also allows Magnetic Engineering to be nimble, enabling us to identify goals and priorities, and then focus on execution. Because we break work up into manageable chunks, we can plan ahead without committing to months of work, leveraging that nimble ability to shift focus to newly identify higher-priority work. On the other hand, we can break larger projects into a series of smaller, more predictable steps, and get feedback from customers as we complete them.

A great example of this is the Magnetic SEM tool currently in alpha release. A full-featured SEM product requires significant effort, but core features useful to alpha customers can be built more quickly. Currently, our engineers are working directly with customers to identify the next set of innovative features to add to the tool. This highlights two more core Agile principles we leverage at Magnetic — interact directly with customers to understand their needs, and build working “thin slices” of applications as early as possible.

As the head of engineering, nothing is more satisfying than seeing our engineers own the development of a tool and drive the expansion of our business. I’m doing my job only to the extent that there are no barriers to the team achieving its two missions of keeping our systems running and continuing to add features to our product that make it more valuable to more customers.

In my next post, I’ll discuss the connection between system architecture and business success at Magnetic.

Cheers,

Mark Weiss

Senior Director, Engineering at Magnetic

by adgenie

Search Retargeting: What do we do and why the new name?

Magnetic(TM) (formerly Domdex) does search re-targeting. Search re-targeting is when we show advertisements to users after they search for particular terms. For example, a user searches for “cell phone” and then goes to another site where we show a cell phone ad.

We think this is interesting given that search ads are the highest converting ads on the internet, perhaps the world. In terms of where we sit relative to other forms of on-line advertising, we think of things in terms of where your potential customers can be found:

We recently changed the name of our company to Magnetic(TM). We had always planned to change the name of the company when we came out of stealth-mode. In order to come up with a new name, we turned to our mission statement. Our mission statement is: We honorably serve the community by helping our customers find their customers with superior efficiency.

We believe search re-targeting today is the most efficient way to find customers because of the large amount of display media, strong indication of intent in search data and other market dynamics. However, we wanted a name that represents us at our core. Magnetic describes the pulling force we provide our customers in finding their customers.

I am proud of what Magnetic has accomplished to date, but we are not perfect and we need your feedback on both our product and the way we think about the market. Thank you very much for taking the time to read my blog and I look forward to the dialogue.

by adgenie

Search Retargeting: Targeting Display Ads Using Data

In 2008, Stephen Baker published a great book called “The Numerati.” The book is a fascinating look at how data is being used in various industries, including healthcare and politics, and opens with an interview with Dave Morgan, the founder of Tacoda, one of the first and largest behavioral targeting companies. In the first pages of the book, Dave speaks about the impressive correlations his team found by looking at thousands of pieces of information per user per month including the fact that people who read romance novels rent cars more frequently on the weekends.

This book, paired with my experiences at DoubleClick and Yahoo!, were the inspiration for creating a company that targets online advertisements using the most abundant and effective form of data online: search data.

Frankly, despite my admiration for Dave and his work, I don’t think he had it quite right. The industry is muddied with companies trying to assign meaning to specific behaviors and the level of detail they engage to examine all of these data points is, at times, overwhelming.

Despite all of the available technologies to enhance digital marketing, and the plethora of opportunities in online advertising, off-line still seems to work much better. This may be a bold claim coming from someone that has founded a search retargeting company, but let’s look at this closer. If you look at my laptop and you look at my girlfriend’s laptop, you’ll see basically the same emails and the same display advertisements.

However, if you look in the mailbox at my house and then in my girlfriend’s mailbox, you’ll find that we’re being served totally different catalogues and direct mail. Not only does the content we receive offline differ more significantly between each individual, it actually maps to the customer’s current stage of life. This means that as shopping habits evolve from Abercrombie & Fitch to J. Crew to Banana Republic to Diesel to Prada, so, too, do the mailers we receive at our homes.

Offline targeting, which in its most basic form simply sends catalogues to past customers or past customers of related companies, works amazingly well. And, at the new data exchanges, we are following this same approach—combining the most basic and reliable online information (search data) to target in-market shoppers or interested potential clients.

The new targeting methods simply show ads to people that previously visited your site (site retargeting) or show ads to people that previously searched for a product (data targeting). This is a departure from the previous generation of behavioral targeting companies that look at thousands of pieces of data per user per month.

We keep it simple. We show car rental ads to people that previously searched for “car rental.”