Payroll part 1: where the software industry began.

Bonus. Arguably the first commercial software demo.

leo the first business computer

(reposted with edits and date corrections from my earlier blog)

Regular readers of this blog, my academic research, or dinner party victims may well be aware of my fascination with the early history of payroll. If not, you will be now.

This post will take you to the dawn of enterprise software. While the history of the software industry makes for an interesting, if odd, hobby, it also helps inform how I think about the software industry today. The past serves as a useful, if sometimes cracked, looking glass. So here goes.

In future posts I will look at where payroll software could be heading.  We are at an interesting point in the payroll industry, with lots of consolidation going on, as well as some promising new developments.

Going back in time

Payroll was the earliest significant business application (beyond a basic inventory management deployment). Lyons Tea Room was a very successful retail hospitality business; consider it to be the Starbucks of its time. In 1949, Lyons Tea Room kicked off a project to build the first business computer, Lyons Electronic Office (LEO).

By February 9th (1951) the payroll programme was completed and first ran live on the 15th for the weekly payroll run on the Thursday and payment to the employees on Friday. (more details here). Despite the inherent unreliability of the machine, a stringent regime of testing and maintenance meant that no one was ever paid late.

An excerpt from the documentation should be recognizable by any payroll person today.

via the archive.

Soon afterwards LEO began to run the payroll for Lyons’ workers, and “despite the inherent unreliability of the machine, a stringent regime of testing and maintenance meant that no one was ever paid late.” LEO then went on to process the payroll for over 20,000 employees at Ford’s Dagenham plant a couple of years later, so this was probably the first outsourced computerised payroll.  In 1954 they spun it off into a seperate company.

One of the first programmers on the LEO was Mary Coombs (nee Blood). Read her obituary here.  The Guardian notes,

The program, written in a simple alphanumeric code, had to produce payslips in pre-decimal currency, taking into account tax deductions, loans to the employee, holiday and sick pay, and calculating the denominations of cash needed to go into pay packets. Coombs later did the same for the Ford motor company, one of many queueing up to use Leo.

Her description of an early bug will bring a smile to any programmer.

I can remember one particularly long evening when it kept going wrong and we were there all evening, because you had to have a programmer involved in this, the engineers couldn’t do it on their own, she recalled. And we eventually discovered that the management lift which went up to the fifth floor where the boardroom etc, was, was interfering. 

Management has been interfering with code ever since. 

Mary was remarkable. 

The first demo?

While researching LEO I came across this document, which is probably the first invitation to a corporate software demo.

We hope that in about a fortnight's time it will be possible to demonstrate the calculator at work. In saying this you will appreciate that, with such a complicated mechanism, we cannot as yet be out of our teething troubles so that it is not possible to be certain in advance that a demonstration will go according to plan.

See the original document here.

The first product marketing brochure is here.

Thompson, one of the leaders at Lyon, had this to say about the LEO payroll a couple of years after the go-live.

I must not, however, give the impression that to put a payroll onto a computer is an easy matter: it is not. The amount of detail to be considered is very large and the number of exceptional circumstances to be dealt with is beyond belief except by someone who has been concerned with a payroll application.

The methods of keeping the records for time worked and other pay entitlements differ very much from one organisation to another, often for no obvious reason. Ideas differ also on what should appear on an employee's pay slip, and the way the information should appear on the payslip is very much a matter of taste. (Thompson, 1958)

What he wrote in 1958 could have been written this morning.

What is remarkable to me is that the Lyons payroll project took less than three years from conception to go-live. This included figuring out the payroll and time rules, inventing a programming language, and actually building the computer.  Of course the tax code is more complex today, but gosh this was remarkable innovation at speed.

You can read more about the history of LEO here. 

All the challenges of change management, executive meddling, middle management intrigue are in the minutes. It reads like any project today, just written with a formal elegance. Spending a few hours in the LEO archive will give you a discombobulated sense of massive progress and almost no progress at the same time.

It’s a paradox that payroll has been relatively stable since 1958, and yet it is always the harbinger of innovation. Whenever I see a payroll pitch, or read about the latest feature enhancements from incumbent vendors, I think back to the early pioneers.

Have a watch of this.

 In a future post, I’ll jump forward 70 years or so, and address what’s ahead for payroll in the context of new work, tax, low code/no code  and business model innovation. (I need to rework the post I did a couple of years ago).

In the meantime, if you are building awesome payroll technology, tell me more about it, please.

I’m of the firm conviction that understanding the history of the software industry gives us a better chance to understand where it may be going. I should probably turn this into a talk at some point. We should be celebrating Mary and her colleagues like we do the Wright Brothers.