Ctrl + C => Ctrl + V is a language that is deeply ingrained to my fingers. They allowed the elixir of life to take a dusty spreadsheet system and revive it, no automation necessary.
You likely know what I’m talking about– tabs and tabs of pivots, reports, forecasts and assumptions all laid out in a neat format. For the everyday spreadsheet poweruser, you would have likely performed these keystrokes hundreds or thousands of times over the course of a career. They allow an amazing, yet static system to stay up to date with the latest data so the output is relevant to the users and stakeholders of the organization.
The beautiful thing with spreadsheets is how it enables problem solvers of any technical ability to get the job done. This is also how I started my career in analytics at a major financial institution that was the first to apply “data-driven insights” into their core lending strategy.
I want to take you back to my early days at Capital One, where a typical project I worked on would last 2 to 3 months. The first 2-3 weeks was where the bulk of the spreadsheet system creation was performed. For a typical project in the world of credit risk, this analysis would consist of getting the latest loan chargeoff performance of past originations by risk tier to update valuation forecasts.
The remaining 6-9 weeks were spent creating a recommendation, collaborating with stakeholders and iterating. At the start of each week in this phase, I would pull down the latest risk data: delinquencies, charge-offs and recoveries, and perform the Copy & Paste maneuver to update the data in my project spreadsheet system.
I never thought too much about this. Mostly because the process worked. That is, until one day I made a huge mistake that almost cost me my job.
A day before a high-stakes meeting with a senior Credit Risk Manager, I had hastily repeated the Copy & Paste exercise yet again to update my presentation with up-to-date data.
But I made the mistake of not updating the bounds of my pivot table. The latest data was in the "data" tab, but the analyses tabs were missing the bottom 20% of records that were sorted by loan default risk.
In practical terms, that meant that my presentation was showing default rates being heavily skewed towards an artificially lower number.
As I proudly presented my findings, I could see the CRM's face quickly shift from a smile to a frown. He of course was a veteran Excel user himself and had a close hand on the pulse of default rates in Subprime lending.
He immediately questioned my numbers, told me to take another look, stood up and walked out of the room. Within minutes of scrambling through my spreadsheet, I found he was right.
A simple oversight like this if left unnoticed could have resulted in a business decision that would have had a dire consequence. Catching it this early was truly best-case scenario, but it still cost my credibility and my ego took a hit.
Fortunately, I did not lose my job. Capital One prioritized the development of human capital that I have always admired where mistakes were seen as a lesson and a stepping stone towards a deeper understanding of the business we were in. (There was a running joke that in order to become a Senior Vice President, you’d have to make a mistake that cost the company at least $1 million, because every leader had lived and learned through such a mistake).
But I did start to think— there’s got to be a better way to manage the lifeblood of how we propose changes to our credit risk policy.
When I shared this story to a peer in Tech, I got a chuckle paired with an eye roll, and was quickly introduced to the two key tenets of software engineering: version control and automation.
Have you ever seen a spreadsheet system that had a clever naming system to maintain versions?
I’ve seen a lot of different versions of these. The purpose of course is to create checkpoints in a spreadsheet system as users go through multiple iterations, especially when there is collaboration with other team members.
It’s a very intuitive system, and in fact, it's where Tech started as well. It wasn’t until 2005 that a system was invented to effectively manage versions of a codebase. This system of version control is called git. It was widely adopted over the following decade, and is the current defacto version control system today.
There are visual representations of git with tools like GitHub and GitLab that allow you to see which user made specific changes to specific files within a codebase, and allows users to merge two versions of the same file together, all while handling conflicts where the same part of code was changed in a file. There’s more protection around publishing changes to the main version, increased documentation into the changes that have been made, and best of all, it encourages more collaboration amongst developers.
Yet, git was never been extended to support changes within a spreadsheet, and thus wasn’t supported beyond programming. Cloud-based instances of spreadsheets (including Microsoft 365 and Google Sheets) have implemented a “Version History” feature, but they still don’t enable the full benefits of git: there’s zero documentation of what changes were made and users who duplicate a spreadsheet to test a separate version can’t easily incorporate changes back to the main spreadsheet system thus inhibiting collaboration.
The core benefit of programming is the ability to complete a repetitive task by producing a system and setting it on auto-pilot. Not AI or anything crazy, just a set of tasks that are triggered on a basic schedule: “at 6am every weekday: update the data in the spreadsheet system and ensure the pivot tables are referencing the new size of the data table”.
The benefits extend well beyond simply saving time. I’d argue it really is more about quality.
When you set up a repeatable system, you can define the actions that need to happen, test them, and iterate into perfection. The system will then run every time with consistency and complete flawlessness.
In practical terms, instead of data being refreshed in the spreadsheet system weekly, it can be updated daily or even hourly. Instead of rushing to update numbers right before a meeting, the user can calmly focus on the latest trends and insights shown in the report, and strengthen their presentation.
Of course, who wouldn’t want to perform at the technical level of a programmer. It would be a no-brainer to adopt version control and automation if it were available. The hurdle has always been that most people aren’t actually programmers.
Is that it though? Do we just settle with Copy & Pasting to manually provide life support to analytic projects?
No way.
In the world of AI, it is possible to “vibe code” a solution using Python and not actually know how the code works. Admittedly not my best proposal, but it is arguably a better “good enough” version than manual copy and pasting.
But I would push one step further.
Use Go Fig.
Okay, I know you know, when it comes to all the data tools out there, I’m going to be biased towards Go Fig.
But that’s because we built Go Fig specifically to offer business users the ability to have full control of code-powered analytics without requiring technical experience in a way that no other data tool has cared to. Despite all the fancy business intelligence software that has been developed over the years, business users are still defaulting to the spreadsheet because data tools come with a lot of technical barriers to modify and customize.
I can’t begin to tell you about how many calls we have with tech leaders every week who vent about how their business teams are still primarily reliant on the spreadsheet, despite all their effort on building dashboards, and continue to be burdened by the same issues I faced in my early career.
The special thing about Go Fig that no other data tool has been able to solve for, is that it combines the benefits of code with the familiarity of spreadsheet formulas, so that problem solvers who do not want to learn programming can still benefit from the power of SQL & Python.
All with a minimal technical barrier to entry.
Go Fig is also loved by heads of Engineering and IT– by offering a robust and intuitive tool that integrates with their data stack to business users who can self-serve, it reduces those users’ dependence on IT, thus freeing up bandwidth to work on more impactful technical strategy.
Try Go Fig risk-free now or request a demo to learn more details.
