What is a fair price of VA today from your point of view? Don't you think that in the nearest future VA without a solution (i.e. without deep integration with some kind of end user application for exact purpose(s)) will be out of market?
End user customers are open to the idea of using VA to solve problems, but the value to them is provided by a holistic solution that solves their problem. And those problems are rarely articulated in terms of discrete VA algorithms. For example, military customer wants solution to help them keep their base parameter secure. They are sold a solution of video camera and 'tripwire' algorithm. But when the bad guys wear camouflage clothing, they pass through the parameter unnoticed. What the customer might have needed was (say) thermal cameras and tripwire algorithm designed for thermal. The solution to their problem ("parameter security") was the composition of multiple technologies, and VA was just a component of that. The entire solution was not VA in and of itself. So while the customer may be willing to pay big money for a solution, and that solution's price may be large enough to absorb an expensive VA license, it needs to be an entire solution that actually works, and the VA license cost needs to be in-line with the solution cost.
The other thing at play is the idea that software wants to be free. Given that there is no (ongoing) manufacturing cost basis behind software, any given class of software tends to get cheaper over time. So, yes, 30-50$ per channel is much more reasonable than $500-1000, and yes, the VA needs to be included in a solution that provides actual value, it is not sold on its own merits.
I think that your question is not specific just to video analytics.
For many analytics companies, the analytics product is their only product. However, the company still needs a building to work from, payroll people, they need to attend trade shows and do marketing programs, hire support people, and other overhead costs. If the only product they ship is their software, the licenses have to be high enough to cover all these costs, and the costs of the developers, and enough to make a profit.
For a company that offers analytics as part of a larger product line, they can generally absorb some of the fixed costs of business into other products, allowing them to sell the analtyics component cheaper than a stand alone company. There is also volume of business, as John has pointed out in the past, analytics has never been a HUGE seller. So the analytics-only company has two problems- overhead, and low volume. This causes the per-channel license to be higher.
There is also another interesting thing when pricing a product. Generally speaking, the higher the price, the more the customer expects for performance. If you sell something like "Analytics Tripwire" for $500/channel license the customer expects excellent results (even if they set it up wrong). If you sell "Analytics Tripwire" for $50 and it doesn't work perfect all the time the customer is more forgiving. This lowers your support costs and your sales costs. When you ask "does low price make a difference", I think there is data to show that it does.
Having sold various levels of analytics from a hundred dollars a channel to thousands a channel I can say the expectations are the same, it's just easier as a manufacturer to leave the integrator holding the bag by refunding the small amount invested in the "analytics upgrade" and not the whole solution.
"Taking into the account that they are selling complete solution, a price on VA can be any. So, it is a bit cunning."
And force out all the analytic only providers? :)
"What is a fair price of VA today from your point of view?"
Whatever the market will bear... The cost of doing analytics right is high, so either developers need to charge a premium for it (as they have in the $500 to $1,000+ range per camera) or they make their money by bundling it with solutions (the Avigilon approach).
Much like it was presented above, VA is only a part of the solution. VA by itself more often has little value without the proper software front end to present it to the customer in a usable fashion. Some can take a solution and only use the event triggers to gain value, but few end users see enough value in that. Depending on the end user needs, they may need some or all of the following: - rules engine that combines multiple data points or triggers - the dashboard that presents usable metrics for the business - UI that incorporates VA as an overlay with the video - search engine that allows for ease of use - forensic search that leverages VA metadata of complete video scene - reporting that provides the detailed overview - macro to micro drill downs metrics - ease of setup for VA for changing needs - ability for edge based analytics to limit traffic - ability for server or cloud analytics to process centrally - many more not listed In short, there is no one provider that will be able to solve the needs of every application and use case for the end user. Now with a large portfolio of VA and a solution that can absorb it, a solution provider may evolve to the VA channel leader in the market. That does not mean they will be the only VA provider or solution to absorb it and price will be defined by the total solution and the market acceptance for a said vertical or application.
Hmm- I was actually doing a pilot back then using both AXIS and Video IQ (pre-Avigilon). Both were causing high quantity of alarms due to lighting issues (car lights coming up highway) Although the AXIS was certainly more than the Video IQ. After involving tech support on both sides was when I was told about the pixel based.
I had to go to a 2 zone server based analytic to eliminate the alarms from both. But I am willing to take another look. In my office first :). That last little fiasco was a month of pain.