Blind spots of Predictive Business Software

Everyone is aware of the predictive/AI transformation underway in enterprise business applications. The analysts, software vendors and thought leaders have used multiple frameworks and maturity models to explain the onslaught of this wave. In simplistic terms, it is basically a transition (albeit gradual) from Descriptive to Diagnostic to Predictive to Prescriptive to Cognitive. In other words, applications have moved/are moving from “What has happened?” to “Why did it happen?” to “What will happen?” to “What should I do then?” to “How can the app address all these questions automatically?”. Although this model has been primarily used in the context of analytics, the boundary between analytics and operational/transactional applications is fading away. IDC predicts that by 2020 half of all analytics would be prescriptive built on cognitive platform and that these cognitive capabilities will be available on all new enterprise applications.

Software vendors have already started riding this wave. According to CB Insights, in 2016, VCs invested around $5 billion in 600+ companies in AI/Predictive space, at a 60% increase from the prior year. Vendors of all sizes including Salesforce/Oracles of the world to mid-sized companies to startups have been investing, building, hiring and expanding into this space. Not surprisingly, the AI/Predictive enterprise software vendors have been on a high-speed exciting ride to capture the mindshare and the users with their amazing predictive applications.

However, based on multiple interactions with current/potential predictive software users across industries in the US over the last few months, I see a few blind spots glaring. The relevancy of the blind spot, the solutions available to address it and the strategy for the same will vary based on the industry, the predictive business need, the stage of the software in the adoption curve and many more factors. Nevertheless, I have listed below a few blind spots that come up frequently:

Blind spot 1: Variation in Data Sources:

It is obvious that the success of predictive/cognitive software platform hinges on the availability of right data and in turn the right data source. A software provider can make multiple assumptions on the data sources while building the predictive capability. A mismatch in data sources between what the product needs out of the box and what the customer has could negatively impact Time to value, Customer Sat, ROI and a few other metrics.

Blind spot 2: Data Governance:

The term “Governance” may sound bureaucratic but its importance manifests if your predictive app relies on data from disparate systems spanning organizational silos. In simple terms, Data Governance is about policies, processes and a framework to manage enterprise data assets. This ensures that high quality data that your predictive app needs is available and accessible at the right time along with required privacy and security. This is all the more important in industries such as Healthcare and Financial Services.

Blind spot 3: Decision Governance

This is very similar to data governance but for decisions. At the end of the day, a predictive or cognitive application is a decisioning tool. The tool could recommend tactical decisions or strategic decisions. Irrespective, if the customer is not clear as to who is accountable for the decision, who needs to be notified of the decision, who needs to monitor the decision and who is responsible for course correction, the organization could face challenges when things go wrong. A cognitive software vendor needs to be cognizant of the decisioning mindset within the enterprise be it gut-based or data-informed or data-driven.

Blind spot 4: Integration

Integrations have always been a key factor in enterprise software purchase decisions. If you were following the Marketing Technology space, you might have come across Cisco’s MarTech stack here, featuring 39 different applications. It is virtually impossible for a vendor to land and expand with this stack without some investment in integration or its enablement.

With Predictive/Cognitive applications, its importance has increased. One of the key value propositions of predictive applications is that it enables right decisions at the right time. In other words, the ability of the organization to adapt to this decision is time-critical. This entails an integration with the ecosystem so the customer can realize the full potential of the predictive application.

Blind spot 5: Trust, Transparency & Credibility

It is human psychology and tendency to critically analyze or question anything imposed upon. More so, when an external software app makes a decision for them on top of letting them be accountable for the decision. Predictive applications that are not perceived as trustworthy or credible are at a severe disadvantage with respect to user adoption.

Blind spot 6: Slow adoption ramp-up

The predictive/cognitive applications require significant changes in user behavior. This in itself is a big barrier to cross.

On top of it, there are solutions that are pitched as an entirely new marketing concept (like ABM in the B2B marketing landscape, as an example) where I noticed significant end-user education is paramount to minimize the confusion in the market. If your predictive capability sits on top of such new paradigms/concepts then the vendor should plan to cross multiple barriers before planning to cross the chasm, say.

In either case, it is worthwhile to consider one use case or a subset of use cases (with some creative product packaging and pricing) as a starting point before letting the customer adopt the full set of predictive capabilities.

In summary, these are just a few of the issues or concerns that predictive application buyers/users raise. The options to fix or address these blind spots vary depending on the industry, user persona, buyer persona, stage in the technology adoption curve, and more importantly the predictive/cognitive application context. (They are topics for an eBook or War Stories possibly.) These blind spots are relevant and applicable irrespective of whether you are moving across the technology adoption curve or crossing the chasm, or if you are doing customer discovery or customer validation for your startup. As a product manager/product head or a sales/marketing/customer success professional, your awareness of these blind spots would go a long way in ensuring your high speed predictive/cognitive ride stays on track.