close up of Excel spreadsheet on a computer
Feature

The Problem With Adding AI to Excel

4 minute read
Erika Morphy avatar
As exciting as it is to add AI to a productivity tool, the reality is the end results are going to be limited no matter how sophisticated the AI is.

Earlier this year Microsoft announced it would integrate artificial intelligence-based productivity features into Excel. It was an exciting development for many users of the spreadsheet as it promised to make several tasks easier and more intuitive. Other users have noted how these promised features — which are still in beta —could deliver far more productivity if Microsoft continues down the same path of improvement.

One suggestion touched on a conundrum that productivity tools will face as they increasingly integrate with AI. Mark Sami, VP of Microsoft and Cloud Solutions at business consultancy SPR said for AI and machine learning to be useful, data from inside the organization is going to have to be made available in Excel. “If an organization is able to build a model that can recognize that a user entered a SKU in a spreadsheet, for example, and quickly allow the user to pull in data from an ERP, forecasting tool or sales tool, that would be a game-changer,” Sami said.

Related Article: Setting Expectations on Excel's Addition of AI

Excel Wasn't Built For This

Which brings us to the conundrum: that isn't how Excel is used or was designed to be used. While Excel can connect to higher-volume data sources, in practice that’s not how most people use it, said Doug Bordonaro, chief data evangelist at AI-analytics platform provider ThoughtSpot. “Excel is most often used as an endpoint solution for desktop analytics: download some data from somewhere else, then play around with it in Excel.”

Bordonaro noted that while Excel increased the row count in a worksheet to just over 1 million a few years ago, that is still very little data for any kind of AI algorithm to play with. "To really leverage the power of AI, users will need to go to where the data is — in scalable BI platforms — and not bring it to their desktop," he said. This doesn’t mean the power of AI isn’t available to Excel users, Bordonaro continued. "But the primary value will be surfacing features of Excel, not insights from the small amounts of data being analyzed," he said. "Instead of burying advanced functionality three menu layers down, AI will let Excel intelligently surface that capability when and where it’s needed so that even advanced functions are available to more typical users."

Learning Opportunities

Related Article: Easy Solutions to Your Worst Spreadsheet Problems

Other Competitors Have an AI Head Start

Users — and perhaps vendors too — should also consider the purpose-built tools in the market that will likely do a better job than a productivity tool that has been jerry-rigged with AI. In the case of Excel, when Werner Krebs, CEO of internet of things analytics startup Acculation, first read about Microsoft’s plans a year ago, it seemed to him as if Microsoft was largely responding to competitors such as Tableau. "Tableau uses AI to recommend a visual display of data and is aimed at less-technical, less-analytical non-data scientist corporate teams that need to manually tweak a large number of data graphs for presentations," Krebs said. "Traditionally, firms that already own a site-license to Excel would use Excel for a lot of the manual data visualization tasks now partially automated by tools like Tableau, so the mere appearance of this category suggests weakness in Excel."

Besides Tableau, he continued, a number of other software dashboard tools have aggressively added AI as a distinguishing feature where Excel or related products might once have been strong competitors. Some examples include IBM Watson, which works with IBM’s Cognos dashboard, which is a competitor to Microsoft Power BI and Excel. Google Analytics has added AI and some of these features could be considered competitors to Power BI or Excel, Krebs noted.

Perhaps even comparing Excel to these tools is premature, suggested Sumit Bansal, founder of Excel tutorial provider Trump Excel. "AI in Excel is still at a very nascent stage. What Microsoft has introduced is a step in the right direction, but it isn’t going to change the way people work in Excel." Which brings up a whole new problem with adding AI to Excel. Because if it doesn’t change the workflow or productivity level, what is the point?