The Riya Pivot

Riya is a wonderful live study in consumer Internet investing.  Munjal has written extensively about how he launched with one thesis (to much fanfare), but went back to the drawing board to re-invent the business a mere six weeks after the launch, even as traffic was growing. It just wasn't monetizable enough.

This week Forbes has an article (reg. required) on "visual search" that does a cursory review of what the term means and who is doing what to whom. We are fortunate to be mentioned in some detail in the second half of the article. We also had a board meeting at Riya this past week and got a hands on test drive of some of the searching capability they will be releasing. 

There is still a lot of work to do, but I love what I see.  I'm not just referring to the user experience.  I love what I see in the business, the team, and the clarity of purpose.  When we first hatched the idea of Riya in my office in late 2004/early 2005, we didn't anticipate this business at all (Forbes outlines how Riya will be used for product search as well as people search).  My perspective was pretty simple.  Drop a great team with interesting technology into a transforming landscape (digital images), add capital, and watch.  We had some ideas about revenue models.  They seemed plausible at the time, but we really had no idea.  No one did a discounted cash flow model.  (In fact, I've never done a DCF to derive a valuation).

Obviously the story is still in process.  We haven't generated $1 of revenue.  But what is much clearer now than 18 months ago is

• who should pay us
• how much they spend now
• why (from their perspective) they will pay us
• how big the current total addressable market is
• what kinds of markets we should avoid because we do not have a meaningful advantage
• who our competitors are
• how we differentiate on the key purchase criteria
• where we will get users
• why they will use us
• what the value of a user is to us, i.e., what we can spend to acquire them
• what our barriers to entry are (and aren't)
• what our break even point is
• what kind of talent is still missing in the Company and where we should look for them

In other words, the business model is crystal clear.

So the Riya 2.0 is nothing like the Powerpoint we saw in Q1-05.  And it bears no resemblance to the stuff that was on our whiteboard in the Fall of 04.  But this is the nature of early stage consumer.  Change Happens. Iterate. Pivot. Evolve.

The key aspects haven't changed. The core of the business is still image  analysis and  classification.  We now have 14 researchers just on this problem - perhaps the largest image analysis pure research team in the world - and a huge intellectual property portfolio.  The biggest changes are in how and where we apply the technology. Perhaps the best move we have made is to signal to the photo, social networking, and community sites that we are not in their business at all, enabling us to work together with them and exploit some real economic synergies.  It's just as important for the world to know what you're Not as what you Are.

We are in the process of re-defining image search.  The core premise of what we are doing is that there are lots of things humans can't describe well in text, but we "know it when we see it." We aren't so much about searching for images as much as we are about searching with images.   This is really a different kind of search experience. Faces are the most extreme case. Our brains are highly tuned to recognize the most subtle visual dues, but humans can't verbally describe faces with any precision at all (except for the occasional scar or mole.) 

Riya's approach to search is a perfect metaphor for early stage investing.  Eighteen months ago I couldn't describe the business with any real precision.  We had some great ingredients in a field of opportunity. I figured I'd know it when I saw it.  Thanks to the team, I just did. And to complete the parallel, as a VC, I'd like "More Like This."

Pivot.

Comments

[sourabh niyogi]

I had no faith before, having done machine vision in previous life, but the "Searching for images" => "Searching with image" transformation is absolutely great: it is very plausible to believe that an e-retailer can sell N% more stuff by grabbing etsy-like features, powered by riya.

It seems clear that eretailers would demand that image-based "More Like This" be immediately blended with non-image based "More Like This" -- c.f. Aggregate Knowledge in progress -- and fortunately, image processing researchers have no trouble navigating that ship: its just another cue! When that happens, you won't have an "image processing web 2.0 company", but a real winner!

[Andrew Fife]

Peter:

"what our barriers to entry are (and aren't)"

Do you mean "obstacles" to entry? ;)

-Andrew

[Ryan]

If you make image searching available as a desktop product where all the information will stay on your computer you can easily sell for $100 a copy. Please don't keep this a web thing only.

[david sutherland]

I guess I don't get it... I would think you would build core engine(s) and then produce several awesome apps to showcase the engine potential to generate the buzz needed for IPO or selloff. Working so hard on the engine without several awesome showcases...???

My feeling is that people are "wowed" by killer apps not killer engines. The common man has to actually be able to use the product and see the benefit to start showing popular mega-interest.

IMO.

Best wishes.
p.s. I'm greedily wanting to immediately put your technology to use on all sorts of my personal picture archiving!!! Please progress with grace, style and deftness... to speedy implimentations!