A look into the TinEye…

Years ago, the introduction of Google Image search changed the way you looked for pictures online. Before Google Images you were forced to search for web pages, find a result and then hope it had a picture on it that you could use.

Lather. Rinse. Pray. Repeat.
Often you would come up empty handed. It was hard enough when you were looking for a picture that fit your keywords (i.e. trying to find a piece of stock photography etc.)

But what about when you have a specific picture and want to find more information out about it? Where else is it used, who’s is it, has it been photoshopped? The list goes on. How do you describe the picture to Google in a way that would allow you to find its source?

Describing a random image to an image search site using only words is a crap shoot at best. Sure you might luck into the exact keywords that somene else used to describe the picture but there’s still a very good chance that you’ll have to wade through dozens upon dozens of search results to find the needle in the haystack.

Enter TinEye.
TinEye is the latest release from Toronto’s Idée. Idée is easily, and inexplicably, one of Toronto’s best kept tech secrets. Building cutting edge image search and recognition tools for close to ten years partners Leila Boujnane & Paul Bloore have built an impressive company that works with some of the world’s largest media & creative organizations. Their other products, PixID and Piximilar, both caused jaws to drop when demoed and there’s a good chance you’ve unknowingly used their technology in the past (i.e. Digg uses Idée’s technology to watch for duplicate image submissions).

So What is TinEye?
Basically, it’s an image search engine that uses images as it’s query: give it an image and it will tell you where the image appears on the web.

Idee put together this quick video with Amber MacArthur that is a great introduction:

A Real Word Example
The video gives a great introduction but let me take a moment to put the technology in some real world context for you:

A few weeks ago my friend Sean Howard twittered:

With a link to this image:

Now ordinarily this could suck up hours of time querying Google images etc., hoping to land on a page with this image…

Instead I right-clicked on it, selected “Search this image on TinEye” and in the span of mere seconds I had multiple results, the first of which took me to a page that had more information about who created the work.


Total time from Sean asking to me being able to respond was less than five minutes (and it only took that long because I had to look up my Tineye password!)

Right now, their system is indexing just shy of half a BILLION images but that number is steadily growing. TinEye is in limited BETA but I was able to score a few invites for my readers. If you’d like to play with TinEye drop me an email at first.last@gmail.com or message me on twitter (ryancoleman) with your email address and I’ll get you an invite.

BTW: The image is by a team of artists know as “eboy

[IDEA] Mind State Messaging

Far too many weeks ago I had lunch with Sean Howard (a.k.a. “Craphammer“) for lunch – we’d been talking to him, and his team at SpinGlobe, about some Clay Tablet marketing activities and he wanted to share some ideas.

One idea he brought up was the concept of “mind-states” – in a nutshell, trying to identify what state of mind your target is in. It was a new concept to me (thus why I’m CTO instead of CMO) but made perfect sense once he explained it.

We talked about how to visualize the notion and by the end of lunch we had a napkin sketch that consisted of mapping mind-states to messages, basically the idea of targeting each message to the specific mind states of each user.

Tweaking it Further
That night on the GO train home I opened up Illustrator and decided to play with the concept a little further. Over the following days Sean and I shot the illustration, along with comments, back and forth eventually coming up with this variation:


To which Sean simply responded “You’ve got to post this so we can discuss it with more people” – which I’m doing now, many, many weeks later (Sorry Sean :) )

The premise is pretty straightforward. The idea is broken into four general quadrants “Mind States”, “Needs”, “Features/Benefits” and lastly “Messaging”. Each oval represents an item in that theme. Obviously in practical use these ovals would be text or images describing the specific element. I also used size the indicate importance (or, in the case of features, strength/support) – the bigger the oval the bigger or more important the item.

Mind States rooted in Needs


My initial impression (and the bit Sean and I are still debating) was that behind each Mind State (which I at first considered to be an irrational state), there was a rational need or requirement behind it.

I’ve dropped the notion of rational/irrational from the latest version but the notion of a Mind State being rooted in a real Need or Requirement (or vice versa) is still very much there. For example, perhaps the Mind State was “I want that promotion”, the thinking is there’s a requirement or need(s) in the background that would resolve, or contribute to resolving, the mind state. In this case it may be “deliver on sales targets”.

Needs can drive States, States can drive Needs.

Features & Benefits to resolve Needs


This was all fine and dandy, but the next consideration was how mind states and needs related to your product or offering. For the most part it’s hard to link features and benefits directly to a mind state. As far as I can see, no feature I can put into my software will resolve the your mind state of “I want that promotion” but if you can uncover the true need then you can build features or identify benefits that help resolve it. By recognizing that the users mind state is actually driven by a need (deliver on sales targets) we can now see that our “Automated Lead Identifier” and “Motivational Tool-tips” features can help the user achieve their need, by keeping them informed and motivated, which will hopefully resolve their mind-state.

Messaging around Features to speak to Mind States


Because Features don’t typically speak directly to Mind States we need to close the loop with messaging. Messaging should speak to the mind state of the user. By working through the previous relationships we know that “I want that promotion” is resolved by the need to “deliver on targets” which our product helps solve by “automatically identifying new leads”.

If you can craft messaging that speaks to their emotional mind state you have the opportunity to strike a real cord with them, then back it up with true features that have their needs in mind.

Mind-State Messaging in Product Management

The other side effect that came out of this exercise was the realization that this could also be used to work through product management issues. By using items that are scaled (or colour coded etc.) to represent the importance you can quickly get an impression of how your product’s features & benefits stack up. The image below shows how needs can be mapped to features or benefits, and how you can quickly gauge if your product is living up to the needs of your prospects/clients.


In example (1) you can see that the need is tiny, and likely not very important in the grand scheme of things, but look at the strength (and assumedly the amount of effort that’s gone into it) of the feature in comparison. Likewise in (2) a huge need is basically going unfulfilled.

Obviously depending on who your specifically targeting you won’t be able to get a perfect match (3) – in theory you’d have different Mind state maps for each persona you’re dealing with in the sales/marketing cycle – but with this model it still gives you some insight into the holes you may have in your product. Especially if the same imbalance pops up on every model etc.

Anyways, this idea is still in the “half-baked” stage, but Sean and I really wanted to throw it out into the ether to see what others thought of it. I know Sean has actually thrown this into the mix on some pitches and projects over the past few weeks – but I’ll leave him to comment on where it worked/didn’t work.