“When the wind of change rises, some people build walls while others erect windmills”
– Old Chinese saying
We are swamped with Analytics consulting companies these days. Traditional Strategy consulting companies, Technology consulting companies, System integrator, even former BPO companies- all of them have joined the analytics consulting race. The driver behind this is not a secret and not something that I want to waste too many sentences on. It is a well know fact that AI boom has fueled thousands of “AI” related companies and practices. Everyone wants to cash in.
Does so much Analytics consulting demand actually exist ?
From an Analytics consulting perspective, Yes….there is demand, at least for now. Everyone is scrambling to figure out what this AI and Digital hype is all about and what it means for their business. After all, it is the human mentality to not to be left behind. So a demand exists, across companies of all sizes. This in turn has led to proliferation of Analytics consulting companies of all sizes.
How sustainable is the trend ?
Not sustainable at all, considering the pace of changes around us. As AI hype is starting to get towards maturity, there is rapid automation of Data Science happening, unknown to many.
The automation is in the form of plug and play Data Science tools. There are already tools now that take your raw data and do everything, from data processing, cleaning, feature selection to model selection/recommendation – all of which is automated. Though this is primarily for basic to intermediate supervised learning currently, this will extend to Deep learning as well soon. A couple of weeks ago, I was able to create a image detection tool for my son using TensorFlow in a weekend (2 days!). And I do not consider myself a Deep Learning guru at all. So eventually, Deep learning Data Science will also get automated and commoditized.
Hence, all the current demand for Advanced Analytics consulting will start waning down, starting 2023 (you can hold me liable for this exact prediction) and by 2027, developing Data Science algorithms in Python or R will be essentially dead in non-product development scenarios. Confident enough to stand by these predictions.
So what is the secret to survival ?
Secret is: Build Products. AI Based solutions and platforms !
But not like the ones that will kill the mundane “code development” analytics consulting. At this point, in 2019, there are already 50+ self service, semi plug and play AI tools available in the marketplace.
Being associated with a Digital startup accelerator in India, I am familiar with at least 100+ startups, in India alone, that are working on something that relates to this area.
So my estimate is that by 2023, there will be thousands of such tools in the market, being churned out by companies and developers in US, China, India, Israel, Europe etc. Of course, mergers, acquisitions and deaths will happen, but the market will still be flooded with remnants.
The World will need more AI based solutions than ever
Now I will pickup Supply Chain context from this point onward. As processes get digitized, analytics gets automated and the manual resources decline, companies will need tools that bind everything together.
Consider a Smart Warehouse context.
There will be robots of all kinds – stationary, mobile and smart. Data generated by robots, automated vehicles and sensors everywhere will be captured automatically. Now let us say there will be a platform solution in place to bring all of this together. In a single view. So essentially KPIs, metrics, real time data, Visual Digital Twin- all in one platform.
But then what ? How do your team on the floor make sense of all of this ? How does this data get married with data from other systems ? How does all these KPIs and metrics get evaluated in context of metrics from Labor planning and Transportation planning KPIs as well ?
You see- there is a big business need. And you can develop a solution around this business need. But it is not that simple (who said it was going to be simple ?).
A solution, a successful one, needs to fulfill a need. This is where it gets tricky to build an Analytics solution of the 20s. Needs will not only evolve fast, but will also vary among companies and Industries. Pre-packaged, off the shelf solutions will fall flat.
The winning product will be standardized enough to be scaled quickly but customizable enough to mold as per the business needs.
And obviously building that type of product is not going to be simple. You need to walk a very fine line but can definitely be done, given the blend of right process, technology and Data Science resources. This solution will bring in all the analytics being done in different platforms together, and will leverage true AI in a warehouse environment to optimally plan the end to end warehouse operations.
Maybe, MAYBE, in 2030s, we will be able to leverage these solutions as foundations to build solutions that will get us to the dream of “self running warehouses”. But that is not happening in next ten years (sorry!).
The Supply Chain (Warehousing) context was just an example but the key is to start identifying product opportunities at this point, be futuristic about what the product can do and start your solution development journey now so that by the time the demand for “mundane” consulting starts to die, you have solutions to offer that no one else has.