The Case For The First Business Computer

Author: Nick Pelling, 26th March 2002, Kingston University Business School, Surrey, UK

ABSTRACT
The business cases behind the five proposals made to the board of J.Lyons & Co. by Thompson and Standingford in 1947 - which led to the construction of the first business computer [#1] - are analysed, but found to be strategically lacking. Both an alternate reading of the case and some contemporary implications are then developed.

CONTENTS
1. The case context - J.Lyons & Co.
2. The case study - Thompson and Standingford's 1947 report to the Board
3. Was LEO a success or a failure?
4. Contemporary implications

Appendix A: LEO / J. Lyons & Co. / LEO Computers Bibliography
Appendix B: Excerpt from Thompson and Standingford's 1947 report.
Appendix C: Payroll comparison from "Three Case Studies in Automation"

Fig.1: Tea barges at J.Lyons' Greenford factory


1. The case context - J.Lyons & Co.

J. Lyons & Co. was founded in 1886 [#2] by Messrs Salmon, Gluckstein and Lyons to fill the challenging market niche of catering for public exhibitions: Salmon & Gluckstein brought catering expertise, while the extremely well-connected Lyons fronted the operation. [#3]

From the outset, the company's corporate focus was on logistics, its business focus was on management accounting, and its organisational focus was on process optimisation through new technology. [#4]

Over subsequent decades, it expanded both vertically [#5] and geographically, building up a large chain of tea-shops throughout England. [#6] This was achieved as much through a large front-office (the "Nippies" waitresses) as through a large back-office (two additional clerks were required for each new tea-shop).

In 1923, the young mathematician JRM Simmons was hired from Cambridge to apply modern scientific thought to Lyons' operations and accounting systems. Interviewed in 1970, he said:

"In fact I was engaged to try to build up a system of information for the management of the company which would be superior, more sensible, than just depending upon the profit and loss account and such like. …. In this respect the company was already ahead."

However, calculating machines available at that time only worked in decimal arithmetic - but all Lyons' transactions were done in were pounds/shillings/pence. So, for the sake of efficiency, from 1928 all its internal accounting operations were done in a form of decimal currency. [#7]

In 1931, Simmons set up the Systems Research Office within Lyons, one of the first operations management departments in the country: and instituted a culture of continuous process-based improvement throughout the company, with a particular accent on the latest technology.

One example was their innovative "Recordak"-based system: this took customer orders, valued them, photographed them (on what we would now recognise as microfilm), before returning them back to the customer as an invoice. This vastly increased their systemic reliability, while sharply reducing the amount of paperwork generated.

Simmons, in his 1962 book "LEO and the managers", remarked:

"It is significant that the General Management was prepared to accept these revolutionary proposals. The Company, which had for long enjoyed a reputation for being enterprising in its own fields of catering and food manufacture, was evidently equally prepared to experiment in a purely clerical field, even though it would strike across the accepted procedures on the so-called productive side of the business. And the results were encouraging. This simple system was accurate, it saved queries because in eliminating copying we eliminated the very possibilities of error, and it was quick and economical because there was so little to it.

"From this job we learned … that clerical work need not be purely historical, and that it would be much more interesting and effective if it led to action being taken rather than merely to records of action that had been taken."

During the Second World War, Lyons redirected both its logistical and catering skills towards the war effort, manufacturing such diverse things as munitions and "Frood" (a frozen food product Lyons invented). However, the post-war economic depression led to a sharp downturn in the performance of its tea-shops. Clearly, times were changing - but what direction should Lyons take next?


2. The case study - Thompson and Standingford's 1947 report to the Board

In 1947, two well-respected employees (T.R.Thompson and Oliver Standingford) were sent over to the United States by Lyons to seek out any advances in office technology made there during the war years. Specifically, they had seen press hyperbole about the first "Electronic Brains" being built for the US military - but weren't allowed to view anything. [#8] However, while there, they heard about Dr Maurice Wilkes' experimental EDSAC computer back at Cambridge, and became excited about the commercial possibilities that this might afford.

On their return, they quickly met with Wilkes, and drafted their historic report to the Board (see Appendix B), to which Simmons added a covering memorandum.

Technology aside, this report proposed one passive scenario…

…and five active scenarios:

The report also flattered the Board (as to its influence and power) by appealing to "wider" aims, though without specifics as to how these might be achieved: [#10]

This machine may well be a prime factor in relieving the present economic distress of the country. In this latter respect we cannot help but feel that Lyons occupies a key position…

As for the financial argument: the machine was predicted to cost £100,000, and to provide cost savings of £50,000 per year - though neither figure was explicitly justified.

The Board's decision? It initially decided to back the Cambridge group until such time as a working demo was produced, but readied a team to design and build it in-house to their own specifications.

Then in 1949, when Wilkes' demo finally happened, the Board fully green-lighted the go-it-alone option, which soon led to LEO I… the world's first business computer.

Fig 2: LEO 1


3. Was LEO a success or a failure?

Operationally and technologically, LEO clearly fulfilled its design aims - payroll calculations that previously took a clerk (using adding machines) eight minutes to perform now took 1.5 seconds or less - a staggering 31900% faster.

However, from the corporate and business points of view, it is hard to see what Lyons gained. For example, according to one source LEO never caused any redundancies there. [#11]

Before LEO, Lyons already had an industry-leading (though admittedly clerk-centric) Information System, that gave it near-real-time management information on more or less all aspects of its business. It had no competitor even close to its standards of quality monitoring and operational efficiency, and had fifty years continuous experience of operational innovation and optimisation… so what competitive advantage did it have to gain?

As for LEO's economics, Bird notes (p.169):

An operating loss had been made on six of the eleven LEO II computers manufactured between 1957 and 1961 and subvention payments to the value of £430,000 from other Lyons' subsidiaries became necessary. … By 1962 Leo Computer Ltd's bank overdraft was over £1 million with an additional £1 million owed to other creditors.

The main reason for this under performance was the difference in price quoted for machines and the actual costs ex-factory, and the apparent inability to reconcile these. This may have been due to the lack of a proper accounting function, for it was not until 20 June 1960 that the first accountant was appointed. Another significant factor was the lack of control or realistic charging rate for their senior consultancy work, much of which was devoted to redesigning their clients' office systems, for which it received little or no revenue.

Many authors feel that what brought on LEO's eventual downfall was the assault on the market by sales-driven computer companies, whose salespeople made whatever wild promises were necessary to secure a deal. An empathy-driven company like Lyons, whose interest was mainly in trying to understand the "true requirements" [#12] of a company before trying to craft a system around them, would always find it difficult to compete against that.

However, to the present author, it seems as though Lyons - both company and board - took its collective eye off the ball in 1947 and lost sight of it for the next 15 years.

Despite having industry-leading operational efficiency, Lyons was, in 1947, heavily exposed to a competitive, commoditised [#13] and shrinking market, at a time of great economic hardship and societal change. These would all independently seem to indicate caution and/or diversification - yet it decided to press forward with a high-risk operations-driven IT installation all the same.

Like many other companies post-War, Lyons was simply caught in a strategy vacuum - in its case, this merely amounted to continue ducking the same fundamental business challenges which it had managed to avoid facing up to (over the previous decades) through growth:

The obvious conclusion to be drawn from all this is that for the post-war Lyons, IT was little more than a vast displacement activity, an operational strategy in the absence of a corporate or business strategy. Lyons entered the computer business with no satisfactory business outcome in mind - so the fact that none did happen should be no surprise.

LEO's business failure was, then, clearly signposted even before its commencement.


4. Contemporary implications

From the case context, it is clear that by appropriate use of new technology, Lyons had (even by 1923) already developed a sophisticated Information System. To the degree this was intended to produce forward-looking information, it definitely foreshadowed much modern MIS thinking. [#14]

Also, in the way that Lyons had a policy of continual operational improvement through the 1920s and 1930s, it would seem to have anticipated kaizen by several decades.

Its continuous metrics, responsiveness to customer-related problems, and iterative management also have quite a modern TQM feel to them.

For modern companies, this points to these three movements' being technology-independent. Their underlying ideas are not novel, but are instead deeply rooted in the kinds of scientific management concepts first espoused in the 1920s, as deployed by Lyons here.

In addition, the introduction of LEO closely matches Hammer & Champy's (1993) general template for a BPR event, in that it not only had board-level support, but was also certainly:

Land (2000) explicitly compares LEO Computers' pre-sales systems analysis with BPR [#15] - but this is slightly unfair on BPR. For all its technological progressiveness, the Lyons' worldview of process optimisation was essentially static, if not backward-looking - in instances where its systems analysts instinctively wanted to recommend a major process redesign (Dunlop, for instance), they pulled up short.

However, even with fifty years of operational experience and a deep understanding of its own business, Lyons was not able to perform a satisfactory BPR transition on itself - because the nature of its business was in flux, and it attempted to re-engineer around its past, not around its future. Strategically, BPR is best executed as a future-directed activity.

Yet even while LEO was being developed, Lyons began - partly to reduce its daily mountain of paperwork - to transform its waitress-service tea-shops into self-service cafeterias, commoditising its core business even further. This was arguably even more of a fundamental BPR event than LEO.

For modern companies, then, the introduction of process innovation through IT should emphatically not mark the end of process innovation by other means.

Ultimately, the main strategic thing Lyons wanted from LEO - improved management information - never really materialised, perhaps because of its technical constraints. [#16] It is hard not to speculate that Lyons hoped (in vain) that the information thus gained might somehow deliver it from its strategy vacuum.

It is easy to find examples of contemporary companies that, caught in a paralysis of excess strategic choice, similarly either prevaricate or choose some inappropriate MIS path to try to resolve the deadlock. However, answers to such strategic dilemmas are normally best found through analyses of core competences and future value than through technological fixes.


Appendix A. LEO / J. Lyons & Co. / LEO Computers Bibliography

Books

Bird, Peter, (1994), Leo: The First Business Computer. Hasler Publishing Limited (Wokingham, Britain).

Caminer, D.T., Aris, J.B.B., Hermon, P.M.R., Land, F.F., (1996), User-Driven Innovation: The World's First Business Computer, McGraw-Hill, Maidenhead. (UK edition)

Caminer, D.T., Aris, J.B.B., Hermon, P.M.R., Land, F.F., (1998), LEO: The Incredible Story of the World's First Business Computer, McGraw-Hill, New York. (US edition)

Hammer, M., Champy, J., (1993) Reengineering the Corporation, Harper Business.

Owen, Jo, (2002), Management Stripped Bare, Kogan Page, London.

PEP, (1957), Three Case Studies in Automation , PEP, London.

Simmons, JRM, (1962), LEO and the Managers, MacDonald, London.

Papers

Hammer, M (1990) Reengineering Work: Don't Automate, Obliterate, Harvard Business Review July/Aug

Land, F.F. (1998), LEO, the First Business Computer: A Personal Experience, in Glass, R.L., (ed)., In the Beginning: Recollections of Software Pioneers, IEEE Computer Society, Los Alamitos, California.

Land, F. F. (2000), The First Business Computer: A Case Study in User-Driven Innovation, IEEE annals of the History of Computer Science, Summer.

Thompson, T.R. and Standingford, O., (1947), Report on Visit to USA, May/June 1947, archives of J. Lyons and Co.

Web resources & Photos

LEO Computers Society's website

National Archive for the History of Computing's webpage on LEO-related documents

Photo of barges taken from the J.Lyons & Co archive at the London Metropolitan Archive

Photo of LEO I taken from the LSE website

 



Appendix B. Excerpt from Thompson and Standingford's 1947 report.

 

VIII. Steps that might be taken by Lyons to advance the development of electronic machines

Our first concern is, of course, the advantages that Lyons may gain from the commercial development of electronic machines, but there is a wider aspect which cannot be overlooked. This machine may well be a prime factor in relieving the present economic distress of the country. In this latter respect we cannot help but feel that Lyons occupies a key position; no one else here, as far as we can learn, has realised the far-reaching possibilities of electronic machines.

We assume that Lyons will want to take full advantage of these machines for their own offices. It is possible for us to play a passive role by merely keeping in touch with developments, and in due course buying machines as they become available, probably from American sources. But such a role would not enable us to have any influence on the kind of machines built, and without commercial influence they may well be built in a form more suitable to handling mathematical and census calculations owing to the influence of the large governmental concerns.

… If we are to play an active part in developing electronic machine there are a number of alternative ways in which we could act:

    1. We could encourage Professor Hartree to carry his research in directions which will produce machines of high memory capacity suitable for commercial purposes, working with him by providing clerical procedures for his experiments. Further we might support his work financially, or endeavour to influence other bodies such as the Nuffield trust to support him.
    2. We could put our ideas in the hands of a large electrical concern such as E.M.I. or G.E.C. and leave them to exploit the machine
    3. We could work with Electronic Controls Inc., of Philadelphia, who would probably want some assistance in financing their project.
    4. We could approach the British Government, placing the matter in their hands to coordinate research in this country and make all resources available to make Britain first in the field.
    5. We could build a machine in our own workshops, drawing on information and advice from Harvard Universities.

--------------------------------------------------

But note this quotation from Bird (1994) p.37:

"When Wilkes returned from the United States the machine he built, the EDSAC, became one of the world's first working alterable stored program computers. Although he achieved complete success, he has since confirmed that his project had been a crash programme. He had employed simple techniques, avoiding frills and the exploitation of technology. The architectural design and decisions taken were his, including the bad ones, and he made no attempt to make the design more economic or elegant. It was this machine that became the basis of Lyons' electronic office."



Appendix C - hardware analysis taken from an appendix in "Three Cases In Office Automation" (1957), Table 1.

COST OF SPECIMEN COMPUTER INSTALLATION

Basic Computer  
  Computer with 3 input and 2 output channels
£67,137
Ancillary equipment  
  2 card readers
2,600
  2 tape readers
650
  1 printer
6,200
  1 card punch
2,350
   
-------
£11,800
Stand-by ancillary equipment  
  1 card reader
1,300
  1 tape reader
325
  1 printer
6,200
  1 card punch
2,350
   
-------
£10,175
Stabilising and maintenance  
  Power stabilising equipment
1,300
  Reserve power stabilising equipment (generator)
495
  Maintenance equipment
1,216
   
-------
£ 3,571
Data preparation  
  5 perforators at 113
565
  5 comparators at 720
3,600
  5 teleprinters and readers at 400
2,000
  2 power packs at 106
212
   
-------
£ 6,377
   
   
TOTAL
£99,060

 Table 2 (not included here) then calculates the cost per hour, based on a double shift (giving 64 productive hours per week, with clerks earning about £400/year), plus cost of equipment at £100,000 (depreciating at 10 per cent per annum), ie £3 depreciation + £5 salary = £8 per productive hour.

Payroll batch job

Without LEO I

With LEO I

Staff hours

4657

1166

Computer hours

-

6.5 (based on a double shift)

Average cost per payroll employee

12.71d

8.07d

However, the cursory calculation behind this 36% reduction in running costs has several key omissions, which should be immediately apparent: