Stipple -vs.- Thinglink -vs.- Pixazza (Luminate)

The primary differences between the companies lie in four areas:

  1. Accuracy. Information versus Ads.
  2. Scale. Automated versus Manual Tagging
  3. Distribution. The Open Web versus Walled Gardens.
  4. Platform. Online images are an ecosystem of advertisers, audiences, publishers, photo agencies and photographers — is there a solution that benefits all constituents?

ACCURACY

In order to better understand the differences between Stipple, Thinglink and Pixazza (now Luminate), ask yourself the following question: why would I or any web user intentionally mouseover photos?

Do you believe people mouseover photos to see pop-up ads that have nothing to do with the photo, or do people mouseover photos to find out who or what is in the photo?

At Stipple, we believe that the only reason audiences will interact with photos long-term is to perform a tactile search and to discover accurate information on what’s in the photo. Thinglink appears to share our perspective. Does accuracy matter? I suppose it depends on your audience and use case.

According to Pixazza / Luminate, they provide inaccurate ‘get the look’ tags because users cannot “afford” what celebrities are wearing. This obfuscates the actual reason they don’t provide accurate information. Pixazza / Luminate has not built the processes or technology to accurately identify the products, brands, prices, urls of products contained in photos. Pixazza / Lumninate’s in-accurate information leads to examples like this real-life:

main-qimg-5b653256cdb97a94a987cabf4928a1d8

David Jones department store put on an expensive fashion show and hired none other than Victoria’s Secret model Miranda Kerr to present several lines. And it worked, as several leading publications ran stories featuring looks from his show. But on sites that use Pixazza / Luminate, David Jones value was largely destroyed. On those sites, mousing over David Jones’ department store images sends people to JC Penney to buy clothes from the brand Arizona — even though the story is clearly labeled *David Jones Fashion Show*. Obviously this is complete fail for David Jones, publisher and users.

At Stipple, we firmly believe that the act of mousing over a photo is a tactile search and users deserve and expect the same level of accuracy when mousing over a photo that they get when they perform a Google search. So when you mouse over say, Selena Gomez’s shoes, you’re going to find out what those shoes are first. IF those shoes are too expensive for you, THEN Stipple help you find a lower cost option. But before we show you anything else, Stipple is going to answer your question accurately.

main-qimg-62a9c71e2e58d3d10bc932deea636cab

Example:

Imagine for a moment if Google search behaved the way that Pixazza / Luminate and other so-called in-image advertisers perform. You’d search for say:

main-qimg-c67825c9287af93a50cfe67d5bbfb874

When you input that search, you’d expect to see search results that look something like this:

main-qimg-417a3007323f1604cd0a987822dbbaca

But instead of showing you the above results, Google says, “we know you can’t afford these shoes“and makes the determination to show you a search results page that looks like this instead:

main-qimg-290c43506e6628a7f77308d255c8c9d7

If you believe, like we do at Stipple, that tactile search / discovery are the only reasons users will intentionally mouseover images long-term, then solutions like Stipple and Thinglink offer viable solutions to the accuracy problem. If you believe that web users mouseover photos because they want to see link spam, unrelated information or don’t care what pops up, then Pixazza / Lumninate offers a viable solution for your use case.

Scale

Nearly all of the companies in the space variously cite similar statistics about the overwhelming number of images on the web (1 trillion plus) and the volume of images being added to the web each month (10 billion plus). Yet, with the exception of Stipple and Thinglink, none of the companies offer viable solutions to accurately tag the content of a single image, much less millions of images.

Stipple is the only company to address the problem of accurately tagging objects within multiple related images. While many businesses have attempted various solutions to the problems: computer vision, crowd sourcing and self-tagging (i.e. the person with the photo, tags their own photo). Each of these solutions fails for various reasons. Due to the advent in digital photography, albums of related images continues to grow rapidly, yet the ability to quickly tag the products does not exist.

main-qimg-c5b89569dbeb1259c6f33474857d7a53

A night out, a lookbook, product shots or a red carpet event might result in literally hundreds of images that contain the same products. And some number of those images may go on to live on hundreds or thousands of different websites. Without tools to tag all of those images quickly, accurately and painlessly, the sheer volume of images will always outstrip our ability to tag them.

main-qimg-9dcaad7e2bc1ea6bf2c6c81f35f7e200

Is Tom Cruise wearing a burgundy blouse? Pixazza / Luminate fails to tag images accurately altogether. Further, Pixazza / Luminate fails to tag the same products uniformly from one site to another. You can literally find Pixazza / Luminate instances where the same product is tagged differently from one photo to the next.

main-qimg-6a3cf28103c17ac5702a82fed84fe7dc

Thinglink only offers the ability to tag each image by hand, product-by-product, picture-by-picture, like all of the little dots in the above image. Need to tag 500 red carpet photos or 50 family pics from your boating trip last weekend? It’s going to be a long night using ThingLink. And as we’ll discuss in the next section, after you’ve taken the time to tag all 50 pictures by hand, the next website to use any of those pictures will need the same tags added all over again.

Manual, laborious, slow tagging isn’t the way the web works. This also isn’t the way images / events build large online audiences. Different sites might run different images of Nicole Kidman in the same dress over the course of the evening. Using Pixazza / Luminate would route audiences to Macy’s or Target for a knock off and users of Thinglink would be forced to label every image by hand.

main-qimg-b6d1f05a6b77290ba6f957817c4b646d

Tag an image accurately once in Stipple and our technology will automatically tag that object in every other photo that contains that product in our database. We do this by machine, we do this in seconds and we do it accurately. So a single accurate tag spreads itself wherever that product is found in other photos. Users of our technology can tag 5 pictures and cause another 200 images to be accurately tagged with exact product placements. Wherever those 200 pics go, our tags will follow.

DISTRIBUTION

Once you’ve tagged a single image accurately and then tagged all the photos related to it, either by hand or using Stipple, then the final problem presents itself: Images don’t build audiences by sitting on one website. They build audiences by moving throughout the web. Thinglink and Pixazza / Luminate essentially built their products for a the tags to live on a single web site. The moment the image moves to another site, both services degrade quickly. As an example, imagine taking a friend-tagged photo from Facebook and putting it on your Tumblr. Will the friend tag follow the photo? Absolutely not. Our competitors solutions behave just like Facebook tags, they’re stuck on a single site.

main-qimg-5d6829d4c6b44a8b01e66b17d9afada3

Here again Pixazza / Luminate offers no actual solution beyond link spam. Want to read a story about Rihanna in Paris? Pixazza / Luminate will cover up the story to show you a movie preview for Brides Maids on one site and ads for Best Buy on another. Pixazza / Luminate views images as real estate for banner ads and puts whatever they want on top of images, irrespective of context. Even if another site uses Pixazza / Luminate’s javascript and uses exactly the same photo, that photo will contain no information other than link spam until their “crowdsourcers” tag the photo again.

main-qimg-430f70dc011efed668facec6238761a0

Thinglink offers a partial solution to this problem: embed codes. If publishers are willing to break their existing publishing workflows and use the Thinglink embed codes for image publishing, Thinglink can distribute information with a single image from site to site. But this fails for all publishers unwilling / unable to use their embed codes or unwilling to display exactly the same image that appears on other websites. This fails for publishers that want to put the image in a photo gallery. This fails for all publishers that don’t source images on the sites of other publishers. This fails for the image rights holders whose content is being illicitly re-distributed via embed codes.

main-qimg-37db4e60e2990eed60f919d201eb8c8f

Stipple offers the web’s only solution for distributing accurate information on top of photos wherever the images goes, throughout our network. In the above example, if any of the 2,000 publishers in our network uses the photo on their site (without embed codes, flash or other gimmicks) the tags will instantly appear on the image when published, even if the image is photo shopped, resized, turned black & white or put into a photo gallery. Stipple not only ensures that the image tag stays with the image wherever it goes, we persist proper attribution information and ensure that the rights holders, celebrities and publishers earn money as the photo is used commercially throughout the web. Every site using the above image would earn a ratable share of the $22,000 ad campaign in the image.

Stipple does this using web standards and it works in all browsers – web & mobile – even lowly IE 6 without plugins, embed codes or other gimmicks. Tag your photos in Stipple and those tags will stay with it on every site in our network. Publish a Stipple’d photo and earn money as your audience engages. License photos through Stipple and earn money wherever it goes.

PLATFORM

Stipple is the only company to build an image advertising platform that includes the demand, supply and discovery sides of the image ecosystem.

1. Images. We are the only company to actually license editorial images and allow users to upload their own images for tagging and syndication throughout the advertising network. In the 4 months since launch we’ve received over 575,000 editorial images, plus over 200,000 additional sports images and and additional 250,000 plus advertisers’ images in the system. Stipple allows users to upload their images as well as tag editorial images that have been properly licensed. *Neither Pixazza / Luminate or Thinglink properly license any images and enable commercial tags / distribution of those photos in violation of the image license agreements.* Stipple protects the interests of the hard-working creative professionals that produce image content.

main-qimg-8bb710cfcdcdebbad7502aad1afef601

2. Products / Inventory. We allow brands to connect and manage product and inventory feeds in the system. This also enables us to verify accurate products.

main-qimg-f774b425700e0953e33f1d81968863c9

3. Campaigns. We allow brands to manage campaigns in real-time, Ad Words style. Set budgets, time tables, spends, bids, payment methods and media types all from a simple, self-serve web interface.

main-qimg-7c0b2111aefbd0e8c2d3f0d381ffa203

4. Data. Stipple provides brands with real time performance data on their products from multiple concurrent sites.

main-qimg-3883b487d1e425e85ef5dd2048d4f253

Stipple has additional product announcements forthcoming next week, 9/8 which will only serve to further sharpen the distinction between the three service offerings. Essentially we believe that putting ads on photos without accuracy or context is a problem that no one actually wants solved. We believe that people would like to be able to get accurate information from photos and advertisers would like to turn images into points of sale when their products are in those images. We believe that we’ve built the only platform to scably serve those needs, respect the rights of the ecosystem and not spam web users.