Digital technologies


March 26, 2013 by wizeblogmin

The Wize Commerce platform uses big data and statistical methods to drive optimal monetization and ROI for our customers.

It all starts with predictive modeling.

To succeed in a simple system, you gotta have a gimmick (or so they say). But when you’re dealing with complex digital systems—with massive streams of real-time and offline data at every step of the way—you need powerful algorithms to beat the competition.

Like our merchant customers, we use algorithms every day to solve a myriad of challenges, including social interest identification, mobile conversion improvement techniques, SEM keyword selection and bidding, page optimization, product recommendations and ROI projections to optimize cross-channel marketing spend.

But for us, two things are different.

The first is our scale. We create over $1B worth of transactions per year, and have been delivering competitively unbeatable results for over a decade, working with many of the world’s biggest sellers across a huge range of categories.

The second is how much skin we have in the game. The nature of our business is that we have millions of dollars invested in keyword bids and other traffic and revenue optimization techniques at any one time. Essentially, we’ve “bet” the money before we‘ve even begun. So our forecasts better be accurate.

Predictive modeling is at the core of what we do and how we think. Statistical methods drive our technology, which in turn drives our statistical methods… and this virtuous cycle is what drives our business. Over ten years, millions of scenarios, and hundreds of product categories, we’ve developed our algorithms to predict:

  • Quantitative ad value for different sites and publishers
  • Optimal search rankings and recommendations
  • Optimal weightings of social signals at each step of the purchase process
  • Etc.

Behind the scenes, we use Hadoop for data aggregation, and Hive and Cassandra to store and access data across millions of products and user profiles. And we’re always refining. Search algorithm changes last year delivered significant revenue lift on Nextag.com and these algorithms are now used across many of our Shopping Network sites to optimize different objective functions.

Because we’re an independent service provider, when we win, our merchant customers win as well. Here are two common traps we help with:

  • The big search, social, and content ecosystems have created enormous value, but their business models are essentially extractive: merchants who work directly with or optimize only for only one ecosystem are essentially playing blackjack against the house—and the house always wins. The Wize Commerce platform uses predictive modeling to optimize traffic and monetization across all channels, not just one.
  • Over time, individual merchants tend to develop deep expertise in their own data and catalog and the unique quirks of their business. But the danger of narrow specialization is that you optimize the wrong thing, and ignore a trend, signal, or important technique that could help you. We’re expert at such techniques, as our Silhouette, our in-store optimization service, and others demonstrate with real results.

Contact us to learn more about how Wize Commerce predictive modeling expertise can boost your traffic and revenue performance.

The Wize Commerce platform uses big data and statistical methods to drive optimal monetization and ROI for our customers. It all starts with predictive modeling. To succeed in a simple system, you gotta have a gimmick (or so they say). But when you’re dealing with complex digital systems—with massive streams of real-time and offline data…

Displaying 0 Comments


February 27, 2013 by wizeblogmin

Wize Commerce regularly hosts expert talks on topics of interest to our internal global staff, and external partners.

One of our recent talks features experts Jijoe Vurghese, Platform & Infrastructure Manager, and Julian Ong, Principle Engineer UI Platform, talking about the Wize Commerce technology platform.

The edited notes and video for this talk are below.

How we’ve built our Services Platform

Wize Commerce is a platform for acquisition and revenue optimization of e-commerce traffic. We have been profitable for the past 10 years and Nextag is now the #2 paid comparison-shopping site on the Internet.

We’re evolving from a shopping-comparison engine, to offering our Wize Commerce platform to businesses outside of our company. To accomplish this, we have to make it easy for customers to leverage our platform. We think the best way to do this is to make this platform a set of services.

The work at Wize Commerce is progressing along two tracks. First, we are building a framework to support services with features like request prioritization, controlled rate of service requests, and automated generation of documentation to make it easier for people to consume these services. The second track requires developing and deploying the actual services themselves.

Having an SOA (Service Oriented Architecture) helps our partners achieve their goals by having all the heavyweight business logic, such as billing calculation, user personalization, search results customization , etc., go to the back-end of the website. So the front end of the website just has to worry about presentation. 

Some of the services we’ve developed for our shopping comparison sites:

  • Product search service, which allows us to access our vast catalog of millions of products.
  • Our visit and click servicewhich allows us to track users across our site and allows us to bill merchants based on what features the user is seeing.
  • Our Buzz Score service (a Buzz Score is basically a measure of the social buzz around a product), which will help a site create much more compelling social features.

Next on our roadmap is A/B testing services. Over the years, we’ve developed a very robust and highly optimized A/B testing framework in Java.  We’re exposing this experimental framework as a service and creating a Ruby client for it.

How our clients use our Services Platform

When a client wants to sign up for services provided by Wize Commerce, the first thing they do is set up and provision their account. Secondly, they provide their catalog. We support several different formats for the catalog to make it easier for our customers. We call this our import service, which ingests a customer’s catalog and makes it available for search. Then, we agree on the specific products and categories we want to bid on. 

When a user comes to one of our client sites, the first thing that will happen is our visit service will be called. This will create a visit, which will begin the process of tracking the user throughout the site. Next, the client site will call our experimentation service, which will determine which side of the A/B tests the user will be on – for the hundreds of tests we have going on at any given time. The user will then browse around the site for a bit, and when the user makes a significant click, they’ll call out to our click service which will record the click information into our data store.

In front of these three services, we have a shared externalization layer, which handles authentication, throttling, and json translation. What this allows us to do is control access to our services, as well as translate our thrift responses (Apache Thrift is the language in which our services speak) into whatever format the client wants.

Final thoughts

In its end state, the Wize Commerce platform exposes millions of events that occur inside the platform in near real time to our clients, calling our traffic acquisition and revenue optimization services.

Subscribe to the Wize Commerce blog to get news on Wize Commerce expert talks and other events, as we launch them.

Wize Commerce regularly hosts expert talks on topics of interest to our internal global staff, and external partners. One of our recent talks features experts Jijoe Vurghese, Platform & Infrastructure Manager, and Julian Ong, Principle Engineer UI Platform, talking about the Wize Commerce technology platform. The edited notes and video for this talk are below….

Displaying 0 Comments


January 30, 2013 by wizeblogmin

We use agile software development methods across almost all Wize Commerce properties. The product benefits of agile development are well-known, including incremental and immediate value delivery.
 
For us additionally, there is an important psychological benefit: agile methods and systems require our teams to adopt a “real-time” mind set. We look at our systems diagnostically at every moment, scanning a rich data stream to catch symptoms and opportunities early and continuously. Our success, as with most companies, involves finding and exploiting slight advantages. Agile keeps us focused, keeps us learning, and keeps us smart.
 
Not all our merchant customers use agile development for their systems, which of course is fine. But when a rigid product design, development methodology or corporate culture spills over into digital marketing decisions, massive amounts of money are left on the table.
 
That’s where we can be of most help, with our Silhouette revenue and traffic optimization service: helping companies turn on an “agile marketing” approach, regardless of their history or the state of their systems today.
 
We spend millions of dollars weekly, making real-time bids, and figuring out what ecosystems are best for each campaign and keyword.
 
The question we ask with our clients is never, how are you performing compared to your history, or current competition, but:
 
How are you performing compared to your potential?
 
Contact us to discuss how Silhouette can help your business switch to “agile marketing” across every channel and ecosystem.

We use agile software development methods across almost all Wize Commerce properties. The product benefits of agile development are well-known, including incremental and immediate value delivery. For us additionally, there is an important psychological benefit: agile methods and systems require our teams to adopt a “real-time” mind set. We look at our systems diagnostically at every…

Displaying 0 Comments


January 10, 2013 by wizeblogmin

Effectively using the wealth of information at your fingertips is key to growing your e-commerce business.

We recently wrote about how smart e-commerce operators can optimize performance, improve their interface – and ultimately increase their bottom line – by making the most of visitor data. But visitor data is far from the only option.

In this post, we will highlight four ways e-commerce companies can also better leverage third-party data to enhance customer satisfaction and marketing effectiveness and grow their business.

Data Management Platforms

Looking to reach more of the people most likely to become your customers? Data management platforms (DMPs) can help.

Over the last few years, DMPs have become popular with marketers as a way to understand their audience and extend the reach of their campaigns. At their simplest, DMPs aggregate data on the interests, behavior and demographics of users across a network of publishers. They then integrate that information with your company’s own user data to identify the type of users who will convert best on your site – which allows you to then target more similar users across affiliated ad networks.

Demand Forecasting and Dynamic Pricing

Continuous, real-time price updates have become the new competitive reality in the world of e-commerce. And while this move has been embraced by many top online retailers, smaller online merchants are struggling to keep up.

A system that analyzes how price – along with many other factors – affects click-through rates for different merchants selling the same product can effectively forecast demand at different price points. Such a model can, for example, enable merchants to dynamically offer a coupon based on current market conditions that will maximize their total margin for a given product or move a certain number of units over a given period of time.

Wize Commerce also helps retailer’s leverage external pricing and demand data through our Radar price alert platform. Whether through the use of Sillhouette, a widget on your site, or as part of our standard Radar email marketing campaigns, we can help you reach highly-qualified customers who have already expressed interest in buying your products but are waiting for the right price.

High-Quality Content

As a marketer or e-commerce site operator, you know that good content helps attract and convert users. For a retailer, content can be anything above and beyond basic product and price info, such as quality names and descriptions, product reviews, videos, or recommendations – and there are a number of third-party data providers looking to sell this type of content to enrich your commerce site.

At Wize Commerce, not only do we integrate data from third party providers, we have developed our own proprietary content platform, which includes content like:

  • High quality product names and attributes optimized based on consumer shopping behavior
  • Structured product relationships, such as related product models, product lines, etc.
  • Proprietary product ratings and review extracts based on aggregating and analyzing millions of product reviews across the web
  • Personalized product recommendations based on analyzing consumer behavior on our network of shopping sites

Social Data

Nearly every commerce site now has some form of integration with at least one social network like Facebook, Pinterest or Twitter. But many sites simply create buttons to make it easy for users to share their favorite products with friends, and leave it at that.

While this simple level of social integration is often helpful in building awareness and driving traffic, few retailers are taking the next step and leveraging this wealth of social data about their products to improve the shopping experience on their own sites.

At Wize Commerce, we analyze consumer behavior on our global network of high-volume shopping sites to create dynamic, personalized product recommendations. With the right tools, merchants, too, can harness product-level social data from Facebook, Twitter and other sources to enhance the shopping experience for their customers.

Our Radar price alert platform helps our merchant customers use email and onsite notifications to reach highly-qualified customers who have expressed interest in buying your products but are waiting for a lower price

We also help retailers display social context and trends in their product listings (e.g. “#3 most tweeted Television”), incorporate social signals (e.g. number of recent likes and tweets) into search ranking algorithms, and develop onsite and email content pushes of the hottest trending products.

The bottom line is that all e-commerce companies will need to learn how to better utilize and manage third-party audience data if they hope to keep up and stay relevant.

Wize Commerce has extensive experience doing precisely that, and we are continually experimenting with how best to leverage such data to help our merchant partners reach more qualified customers.

So, as you consider how to better leverage this data to grow your business, we encourage you to contact us to start a conversation about how we can help.

Effectively using the wealth of information at your fingertips is key to growing your e-commerce business. We recently wrote about how smart e-commerce operators can optimize performance, improve their interface – and ultimately increase their bottom line – by making the most of visitor data. But visitor data is far from the only option. In…

Displaying 0 Comments


December 16, 2012 by wizeblogmin

Expertly managed Search Engine Marketing (SEM) is essential for any e-commerce company.

It can be summarized into 3 components:

  • Search Keywords: Marketers working at scale must maximize the number of relevant options for every product and campaign. For an enterprise, generating keywords manually would be grossly inefficient and would likely only uncover the most popular, and therefore most expensive, options. Keyword generation must be both automated, with a nuanced understanding of both the short and long tail.
  • Ad Copy: While having a good keyword set is essential, good copy can make a huge difference. In fact, copy CTR (click-through-rate) is one of the elements search engines use to determine how your copy will rank. If you have the right keyword, but wrong copy, your ad might not even get seen.
  • Bidding: Finally, with millions of keywords deployed at any one time, advanced bidding algorithms must be in place. For some high volume keywords, this is relatively straightforward since a great deal of data is available. The challenge comes with the long tail of low volume keywords that usually have very little or no data to work with.

Wize Commerce has been refining our SEM methods since 2002, across all the major search ecosystems.

Here’s how our service benefits you: (more…)

Expertly managed Search Engine Marketing (SEM) is essential for any e-commerce company. It can be summarized into 3 components: Search Keywords: Marketers working at scale must maximize the number of relevant options for every product and campaign. For an enterprise, generating keywords manually would be grossly inefficient and would likely only uncover the most popular, and therefore most…

Displaying 0 Comments