Internal and external indicators point to the fact the industry is officially “crossing the chasm” in the adoption of computer-vision analytics across their networks. It was inevitable, as I have been expressing for years, but there is always that “natural, tech-adoption curve” that needs to be respected.
We have been waiting, building, and working for over half a decade towards the moment where brands, operators, and brick and mortar businesses would realize that they could extract massive amounts of value from having real-time data of what’s happening with their physical world assets. And that data can unleash substantial long-term industry growth.
Today, not only are the small early-adopter companies utilizing this new measurement/engagement solutions but we are seeing that the billion-dollar, publicly traded companies are jumping onboard as well.
So why are these corporations doing it?
Revenue Maximization: the ability to collect real-time data is crucial in changing content intelligently, serving dynamic ads, optimizing A/B testing on-floor merchandise, understanding customer flow, collecting instant dwell time and attention-span figures, and truly grasping visitor demographics in the physical world. Furthermore, feeding this data into ad serving platforms, programmatic engines, POS systems, and yield management platforms enables data to catapult revenues.
Transparency: times have changed. Clients, brands, retailers want real data and the ability to access it instantly. The traditional status-quo narrative on proof-of-performance and measurement is being challenged by clients, who ultimately fund the industry. History-based metrics, one-time surveys, and aggregated sampling statistics with complex formulas to prove viewership still have merit, but nothing can equate to the “truth data” that is now available through the adoption of these new technologies.
Market Demand: Gladwell’s “tipping point” occurred. The word is out. We understood this really happened during the third quarter of 2018. During that period, major players began implementing roadmaps, existing companies went from smaller activations towards full network deployments, and a bombardment of brand related news emerged. Additionally, all the main industry associations are being proactive and welcoming to the vendors of these technological advancements. Today reminisces the period of 1996–98, when most companies suddenly realized that the Internet was real and they could exponentially increase from implementing it; this is happening today with real-time data, measurement, programmatic, and business intelligence metrics.
Owning the Data: clients want to own their performance data and the complete set of viewership and engagement metrics. Applied data unlocks previously unknown intelligence and today’s API’s enables a plethora of opportunities. Companies want to pipe data and results to their other dashboards, they want this flexibility to create models, improve operations, maximize sales, and cater to clients. The granularity and instant collection and analysis of data, outperforms everything that existed previously.
Okay, but what now? How do we go about it?
Let’s assume that a high percentage of readers are responsible for data, metrics, business intelligence, innovation, measurement, and/or revenue management for your company. And let’s assume that from previous blogs, industry news, trade shows and other online research, you decide this is important. You feel you need to adopt this business practice of real-time intelligence before your competitors start gaining market share, or if you are already behind, or if you are forgoing potential high-digit revenue growth, among many other reasons. What now?
The warning for everyone is that despite this technology being one of the main catalysts revolutionizing and evolving our industry, developing the technology stack itself is grippingly complex. This is not for the faint of heart. Only a handful of companies globally provide this technology for the industry and some do it 10x better than others. As a result of these advancements in AI, infrastructure, hardware, and connectivity, corporations are now easily deploying it across their entire networks; you just need to know what to look for in order to succeed and what to stay miles away from.
Artificial Intelligence, machine learning, tagged data sets, training models, and computer vision are highly-focused engineering disciplines. This is deep technology that takes years to build, requiring large top-tier engineering teams working together to ensure a flawless infrastructure and highest accuracy of data metrics. Often, tens of millions of dollars in investment occur before a whole solution can be built. The reason I’m telling you this is to desmistify myths surrounding the adoption of real-time measurement/engagement into your businesses.
People often have the following ideas of how to go about it:
Build It Ourselves:
“This is not so hard, we will build it ourselves, we have smart engineers and corporate backing.”
> Possibly one of the worst mistakes you can do. I have founded technology companies most of my career and can say with utmost certainty that building AdMobilize’s stack was the hardest problem I have ever solved; and my entire engineering team will concur, and they had worked on really tough problems before.
> Don’t attempt it. You will most likely fail, lose millions in investments and opportunity cost. Just work with a provider that has fully anonymous real-time analytics/engagement as their core mission, and hopefully choose a partner with a solid focus serving your respective industry.
> At times we hear that a digital signage company has decided to provide this kind of technology to their clients, claiming they have built it. That should be really worrisome (unless they are using a third-party, that’s acceptable, just find out who is providing the actual tech behind the scenes). How can a company (be it screen manufacturers, CMS providers, billboard operators, media agency, etc) have developed a state of the art computer vision analytics platform in a short period? Will it be any good or sustainable, it’s highly unlikely. For this type of technology to be world-class, a provider has to be solely immersed in it. The analytics company has to be obsessed in solving this problem, not an add-on second-thought feature. It has to be its mission.
Foreign Providers in Countries with Dubious Data Policies:
“I found this company in (fill in the blank country) and they apparently have a super inexpensive solution.”
> Run for the hills please. Many countries in the world have non-existent data protection policies. You don’t want that uncertainty. Also, some countries embrace facial recognition, I would steer away from using providers from those locations. The power of the technology that I’m referring to is “detection,” not recognition. Only implement detection. (If you want to learn more, read here).
> Also you want for your provider to be a registered company, with a real business; a partner that you can trust and hold accountable. So choose providers from countries where the rule of law applies.
> I had a professor at Cornell, who always said jokingly but with a real meaning behind it, “cheap no good, good no cheap.” Be aware of providers that claim to have the cheapest solutions, that should be an alarming sign. There exists face detection only API’s from, presenting themselves as having the total solution. That’s not the case. To deploy this solution at scale ensuring success, you need the whole product, not just an API. I define the “whole product” towards the end of this post.
New Little Company Claiming To Have The Best Solution:
“There is this new start-up that is promising the best ever analytics solution for my (OOH, retail, QSR, etc) industry.”
> I love entrepreneurs, after all, we were a start-up back in 2012 but the reality is that it took us the first 4 years to build our stack, it was really intricate and expensive, and we have been possessed in perfecting it since then. So be careful with claims from a new company that launches, that is guaranteeing to be superior to all others.
> Fortunately or not, these solutions take years and millions to build. I believe that in the next 5 years, there will be a couple more players in the space. It’s a bullish segment and people globally are recognizing that. But it takes arduous effort to launch these solutions.
> Feel free to experiment but don’t risk your reputation and business with unproven players. Work with the best to obtain superior results.
Keeping Large Players In Check:
“I have heard of xyz provider who supposedly has been doing this the longest. I’m going to trust that.”
> This is a tricky one. The fact that a provider has been doing this for a long time has several advantages in terms of being a established company and name recognition. But that does not mean they are the most innovative, have the most complete or accurate platform, or a pulse on the latest tech advancements.
> Very often older companies get complacent, stop pushing the envelope, get behind on what’s cutting edge. So please do your research, due diligence, and work with partners that can continuously elevate and contribute to your roadmap. This is key for the type of solutions we are referring to. In this case, we are talking about AI, computer vision, real-time data, so you need to engage with forward-thinking innovative partners. It can be that those legacy companies have this DNA, often it’s not, but hopefully your perspective is a bit more rounded on what you need to look for.
In line with the above, it’s fundamental to understand the elements involved in deploying these “whole product” solutions. You want to minimize false starts, sunk resources and loss of time; instead, you want total piece of mind in being able to scale global deployments with assurance and confidence.
So, here’s a quick guide of what’s important?
In attempting brevity, I’m not going to elaborate on every point below (most likely on a future piece). For the time being, feel free to reach out to me or anyone on my team and we will be happy to elaborate with you.
Does The Provider’s Platform Include:
- Operating System Agnostic (Windows, Linux, Android)
- Global Device Management (updates, alerts, monitoring, backup)
- Comprehensive Documentation
- Minimal Bandwidth Consumption (both Wi-Fi and 3G)
- Smallest Latency of Data Capture and Analysis (ideally sub 15 milliseconds)
- Highest Accuracy of Algorithms (92–98%)
- Strict Data Protection Policies
- GDPR Compliant
- Media Player Integrations
- Proprietary Machine Learning Engines (not using a third party)
- CMS Integrations
- Advanced Edge-Processing Analytics
- Full Encryption of Data
- Whole Product Under One Platform Including Face, Crowd, Vehicles Analytics (more streamlined provider management)
- 100% Anonymous Data Only
- Size of Trained Databases and Proprietary
- Optimized for Low Power Devices (inexpensive hardware)
- Automatic Backups
- A Full Stack Solution
- Intuitive and Flexible Dashboards
- Easy Data Access and Granular Reporting
- Instant Detection
- Integrations into Programmatic Engines
- Commitment to First-Class Support and Services
- Global Footprint
- Flexibility on Volume Pricing
- Real-time Data Only
- Flawless Data Preservation Hosting
- Ability for Data to Trigger Dynamic Ads Based On Real-Time Audience Attributes
- Patents Held by The Provider
- Responsiveness and Kindness of Team
- Must Be Providers’ Core Focus
- Innovative, Cutting-Edge Culture Ideally (future tech roadmap)
- All-Encompassing API’s for Integrations
I feel that’s a good start. I realize it’s a lot of information provided but I believe this is enough ammunition for you and your corporations to be dangerous (smile) in understanding what steps to take and what to look for.
A few times a year I feel compelled, and honestly a sense of responsibility, to express my views on “pressing” subjects regarding our digital signage industry (involving OOH, retail, communications, smart cities). I try to be very selective when I do emerge to share these observations, as I value everyone’s time/focus coefficient.
If interested, you can find previous thought pieces below on how this type of technology works and what are the forces driving it’s exponential adoption:
Technological progress pushes comfort-level boundaries, forces new ways of thinking, and evolves antiquated processes. Ultimately, innovation creates new revenue streams and business models, increases customer satisfaction and options, and ignites new and fresh industry paths. It’s always a mix of uncertainty and thrills. I wish everyone success in this “new normal,” the journey is poised to be a rewarding one.