Search Predictions for 2018

New Year’s predictions are a longstanding tradition, dating at least back to the 16th century French seer Nostradamus.  Personally, I started only four years ago, and my predictions focus around search, digital workplaces, and market dynamics involving Microsoft.  I intend for these predictions to be much more useful, informative, and accurate than Nostradamus’ (a low bar, I know).

How’s my crystal ball?

Last year I worked to make predictions crisper and more directly measurable. Obviously, none of us knows the future, but looking at how accurate my predictions are helps me improve at the art and lets you readers give me grief when I’m wrong. (Nostradamus has the advantage of being long dead, where I genuinely want to hear your feedback ?).

Predictions from 2017

As you can see from the table below, 7 of 10 predictions came to pass during 2017, for a 70% score.

Jeff’s 2017 Predictions

Did it happen
in 2017?


The Intelligent Intranet will become a commonly-understood thing, and real



1/2 the internet-in-a-box products will die



Teams / Facebook workplace battle will hurt Slack & kill Jive….but Yammer won’t die



Azure versus AWS versus Google Cloud – Azure will gain 10 points



Data Sovereignty will become a factor in over half of enterprise IT projects



Dynamics will gain 5 points on Salesforce



Office Graph will finally cross the “trough of Disillusionment”



Tectonic shift in the web search market



API Mashups will highlight Search



Handsfree & Holographic information exploration will appear


You can read my predictions from last year to get the details.

Some are evident and even obvious in retrospect, though they weren’t at all clear a year ago.

  • Teams and Facebook for Business (#3) – did seriously wound Jive, and they were acquired by PE firm ESQ Capital in June – much faster than most expected.
  • Data Sovereignty (#5) – I predicted it would become a factor in over half (50%) of enterprise IT projects, but with GDPR deadlines nearing, this has been a tsunami-and a quagmire, and the estimates I hear privately from analysts are over 90% (versus 20% one year ago).
  • Office Graph (#7) – after several years where Delve was the primary face of the Office Graph and a transition where it was subsumed in name and function to the Microsoft Graph, many different elements in Office 365 are now powered by the “intelligence” of this fabric.
  • Handsfree & Holographic information (#9) – AR and Conversational Interfaces are now everywhere you look but they were still very directed a year ago and nowhere near the ability to explore information free form.  Now there are many different AR applications that leverage search heavily.

Some of these require a little explanation:

  • The “Intelligent Intranet” becoming mainstream (#1) I had to measure with by asking 10 people what it means, and having 7 of them say essentially the same thing. Back in 2016 I got responses all over the map (Check out my blog on the Intelligent Intranet and the videos from some of these interviews. I received 19 different answers from 27 people. At the end of 2017 I interviewed 20 people, and 16 of them were consistent in what they thought the Intelligent Intranet meant and why you should care….a big shift.
  • API Mashups highlighting search (#9) predicted that the trend towards low-code mashups would start leveraging search explicitly.  My metric was that at least one major mashup contest winner would highlight their use of search and discovery. I know of at least two:
    1. Virtual Librarian won “most useful API mashup” in 2017 – at Amazon Alexa API Mashup Contest
    2. At the first “Actions on Google Developer Challenge” the Credit Card Helper got 2nd overall and top in 3 subcategories

The potential for this kind of mashup is enormous. At BA Insight, we’ve specifically designed our SmartHub product to support this approach, and we’re using it to leverage the cognitive services from Google and Microsoft.

The three predictions that didn’t come to pass in 2017 were all ones where I just set the bar too high too fast. They all showed trends in the direction I predicted, but didn’t meet the criteria I set forth.  I’m carrying all three of them over to 2018.

Predictions for 2018

Here you go: ten predictions for search in 2018. (The three in italics are carried over from 2017 as I still feel strongly about these).

  • Hyper-Personalization takes hold
  • ML becomes essential to search
  • Open Architecture for search gains traction
  • Voice activated search will come inside the firewall
  • Blockchain-based search engines will appear
  • Microsoft will splinter their search offerings
  • 1/2 the intranet-in-a-box products will die
  • Azure versus AWS versus Google Cloud – Azure will gain 5 points
  • Dynamics will gain 5 points on Salesforce
  • Big data will give way to smart data

The first five predictions are around techniques and technologies that will show up in search and discovery applications. The next four are about market dynamics including the “battle for the cloud” and Microsoft’s position. The final one is a megatrend (and countertrend) I see happening.

Prediction 1: Hyper-Personalization takes hold

Personalization has joined faceting as a key technique for effective enterprise search. Over the last 3 years or so, users have started accepting the idea that two different people might get different results based on their role or location – in part because Google has gotten much more aggressive about targeting results and ads in consumer search.

I now predict that this will quickly move to highly granular “hyper-personalization”. This term arose in marketing only two years ago, as segmentation was extended down to individuals. It’s starting to be used in the precision medicine field too. I predict that effective search solutions that tailor results based on individuals’ behavior and characteristics – beyond simply expressing preferences or tying people into groups –  will be in production within the year.

Prediction 2: ML becomes essential to search

Machine Learning is one of the “megatrends” that I usually leave to one of the many general futurist blogs.  However, one of my predictions for 2016 was that “ML will be everywhere”, which seemed unlikely back then…but is indisputable now. Forbes reports that 70% of enterprises expect to implement over the next 12 months. A new Forrester Research report, Predictions 2018: The Honeymoon For AI Is Over, predicts that in 2018 enterprises will finally move beyond the hype to recognize that AI requires hard work.

ML is already in use in search; for example, our image search feature is based on machine learning, as is our entity extraction and several other features.  But you can certainly field a strong search application without these, depending on the scenario.

What I predict for 2018 is that ML will be effective enough and ubiquitous enough that nearly every new search application will leverage it.

Prediction 3: Open Architecture for search gains traction

There’s a saying that “the best way to predict the future is to invent it”. That certainly applies to this prediction (and perhaps a bit to the previous two as well).  BA Insight has been using an open architecture approach, sometimes called our “Best of Breed” approach, for a long time, and in 2017 we doubled down on it by adding in cognitive capabilities using the same approach.  Our customers have told us they get a lot of benefit from the flexibility and power of this approach.

In 2018 I predict that we will no longer be the only vendor with this approach, and that two or more others will have joined us.  Looking forward to the competition!

Prediction 4: Voice activated search will come inside the firewall

In the consumer domain, 60% of online searches are from mobile devices, and 20% of mobile queries are now voice activated. The technology has been getting much better – the word error rate is as low as 8% (compared to over 20% in 2012), and over 2/3 of smartphone users use a digital assistant at least weekly.  SEO pundits are starting to talk about optimizing SEO for voice search. Yet inside the firewall the use of mobile devices and voice activation for search is remarkably low – I haven’t found any published numbers but I chaired a panel on this subject recently and the consensus was that under 5% of searches inside the enterprise come from mobile devices, and almost none of them are voice activated.

I predict that we’ll see a big change this year, and that voice-activated search will be adopted in production applications serving employees within at least 100 organizations. Alexa for Business has just been announced by Amazon, and I wouldn’t be surprised to see a range of specialty applications coming from this (much as we see specialty Alexa skills gaining traction), but some voice-activated searches will be as general as the ones handled by Google Now or Siri.

Prediction 5. Blockchain-based search engines will appear

Blockchain is another “megatrend” that had a breakout year in 2017. They are still mostly associated with cryptocurrencies and Initial coin offerings (ICOs), but blockchain technology itself offers secure distributed transactions, which has some amazing technological potential.

I predict that we’ll see a viable search engine built using blockchain within the year. This may be a highly-secure encrypted search engine for use in intelligence communities, and/or a service that protects your consumer information as an alternative to the major consumer search engines (or both).

Prediction 6. Microsoft will splinter their search offerings

Over the past five years Microsoft’s enterprise search strategy has evolved from a focus on SharePoint to SharePoint Online/O365 and Hybrid and now into several different, and mostly independent, search initiatives. Microsoft says that these are all part of a spectrum of search capabilities that are integrated together. However, there are now four enterprise search offerings from Microsoft:

  • “Classic Search” in SharePoint (on-prem, online, and hybrid)
  • “Modern Search” in Office 365
  • Bing for Business
  • Azure Search

This is a confusing landscape, even without considering the nine different search experiences within “Modern Search” and the embedded search experiences in all the different Office 365, Dynamics 365, and Microsoft 365 services.

2018 will likely see either a rationalization of these offerings or a splintering into independent offerings without the pretense of coordination. Although I am a fan of Microsoft and of Microsoft search, I predict that these offerings will diverge rather than converge. We’ll end 2018 with distinct but overlapping offerings, and with more rather than fewer search experiences.

Prediction 7. 1/2 the intranet-in-a-box products will die

I was wrong when I predicted this for 2017; nevertheless, I’m carrying over the prediction to 2018 and raising the stakes. Starting back in 2016, there was an explosion of new products that built on SharePoint and/or Office 365 to provide a “ready to go” Intranet. At the same time, Microsoft has been steadily progressing an initiative to make Office 365 a really solid intranet out-of-the-box.

Clearbox published a report covering six of these intranet-in-a-box products in January 2016; the 2nd edition came out in November 2016 and covered 26; the 3rd edition came out in December 2017 including 42!  I predicted that at least 13 vendors would pull out of this market by the end of 2017. Only one has visibly done so. It may be that others have thrown in the towel but continue to present a pretty face to the market – after all, if you’ve already got a product built, why wouldn’t you just keep listing it even if it was failing? But I can’t prove that my prediction came true in 2017.

For 2018, I’ll predict that fewer than 20 of these products have healthy sales (which I’ll define as over one new customer per month, eg 12 new customers over 2018 – a pretty low bar).

Prediction 8. Azure versus AWS versus Google Cloud – Azure will gain 5 points year ago, I described the dynamics of the megavendor battle unfolding around cloud platforms, including Microsoft’s departure from their traditional “fast follower” approach. 2017 continued to see a heated battle between AWS, Azure, and Google Cloud.

I predicted that Azure would gain 10 points of market share against AWS – an enormous feat given how dominant Amazon still is in this space. By mid-year, Microsoft had only gained 2 points (Azure gained 3% but AWS gained 1% share). I was incorrect, because I was hugely overoptimistic and didn’t give Amazon enough credit for how competitive they are.

I still see the same basic dynamic at play. At the risk of being wrong again, I’m going to repeat the same prediction for 2018, but modify it to a 5 point gain over the space of 2018. This is by no means a slam dunk – not only is Amazon fighting extremely hard, but Google has upped its game recently and Oracle has just introduced a new “bare metal cloud”.

Prediction 9. Dynamics will gain 5 points on Salesforce

This is the last of my three “carryovers”. I predicted that Dynamics would gain significant ground on Salesforce as they developed Dynamics 365 and integrated LinkedIn. I predicted 5 points of market share within 2017.

Gartner didn’t publish their CRM market share numbers in 2017, nor did any other analyst, so I can’t really say what happened. Dynamics is definitely gaining ground on Salesforce (reversing the trend of the previous two years) but I seriously doubt it was 5 points (perhaps 2).  It was still a great year for Dynamics – it became #2 in ERP. Dynamics 365 is growing at about 70% per quarter, and the LinkedIn integration is just starting to kick in. But it was also a great year for Salesforce, which topped $8.4B, staked out their own position in AI with Salesforce Einstein, and put together a strategic partnership with Google.

For 2018, I’ll predict that Dynamics will gain 5 points of market share on Salesforce. Perhaps I’ll go down in flames on this one – after all, Microsoft has tried several times over the last decade to accelerate traction in the CRM market, and Salesforce is a considerable foe. For Microsoft to gain ground again they’ll need to have all the cylinders working perfectly in their complicated engine.

Prediction 10. Big Data will give way to Smart Data

Big Data has been so overhyped as to become a nearly meaningless term. The hype is mostly groundless, too – as Bloomberg recently pointed out in the article “Big Data works only for those who hype it”. There are certainly some triumphs in the field of Big Data. However, McKinsey says that companies are realizing only a small fraction of their big data-related potential due to a “lack of analytical talent” and “siloed data within companies.”

Whatever Big Data means to you, it’s “big” – meaning very large volumes of data.  But many of the real problems organizations face today don’t really involve trillions of data points; they can be understood and addressed at a smaller scale.  Some people are starting to talk about this differently; the book “Small Data” came out in mid 2016, and the ALLDATA conference – covering Big Data, Small Data, Linked Data and Open Data – is in its third year.

So what do I mean by “Smart Data”? Where Big Data often deals with data that is overwhelming in one of the three “V”s (volume, velocity, or variety), smart data really focuses on value. The orientation with “Smart Data” is to find the subset of your data that actually applies to a given problem, and make that data understandable by smart humans. This certainly means bridging across siloed data within companies. It probably involves many of the same data processing techniques and technologies associated with “Big Data”, such as cognitive computing, AI, and Natural Language Processing. But the data needs to be understood and interpreted in a specific context and approached with a problem in mind.

This is quite a nuanced change, but I think it’s an essential one. To make a crisp prediction, I’ll state it this way: by the end of 2018, the concept of “smart data” will be understood and gaining mindshare among serious practitioners. If Smart Data is largely ‘cute marketing spin off’ of Big Data, this prediction won’t have come to pass. But if I can show you strong evidence that this is truly a different approach and that it is spreading, that should count.

What do YOU think?

I put special attention into making these predictions crisper and more measurable, and hopefully I’ve done so without detracting from them. The real purpose of these predictions is to provoke thoughts, share some useful insights, and perhaps expose you to some things you didn’t know about before.

I’d love your feedback – and your own predictions if you care to make them. Let’s see how these unfold through 2018!