4 Unexpected Costs of Developing your own Plant Breeding System
You’ve come to the conclusion you need a new plant breeding system to manage your data. The options in front of you are to buy “off-the-shelf” from a professional specialist company or develop your own plant breeding system. Which is cheaper? You have some expertise in-house, you also work with a local software company from time-to-time for other small projects. You get quotes from the professional company, however, you think you can do it cheaper yourself using your own staff, whom you are paying anyway, and outsourcing some development to the local company. At face value it seems more cost-effective to do it yourself, after all, you get the system you want and it looks exactly how you want it to. You think it’s straightforward, even common sense, to go it alone.
You were right, Dieter, the software company that we paid over a million dollars to develop our plant breeding system could not do the analytics we required, could not really understand many of the plant breeding concepts and they failed to deliver.
This was the reply from a multinational seed company a few years ago. The seed company had ‘deep pockets’ and some breeders demanded their own plant breeding system instead of licensing ours. There seemed to be some politics between the senior breeders who were somewhat conservative and the younger breeders who were more computer literate and had a “more enlightened” understanding of the issue.
A Plant Breeding System has One Simple Goal
The rationale was temptingly simple – save money by not paying software license fees long-term and get the system that you really want. But as they discovered, that was a gross oversimplification. They had not considered the complexity and hidden costs which eventually “came to haunt them”. This is a mission-critical R&D software system. Did they really want to outsource the development of it to a software company that did not understand plant breeding, experimental design and the multitude of practical user requirements for such software?
Initial Costs of a Plant Breeding System
Despite spending far more dollars than the cost of licensing our software, the seed company’s new system never materialized. If you are tempted to develop your own plant breeding software system, here are the inevitable considerations.
1. Software Design Specifications
First, what are the software design specifications? You need to define exactly what these are. This means time taken from the breeders in the dialogue back and forth. Even if you plan to develop it in-house, you need to go into considerable detail so software development is done correctly and in the shortest time frame. This means the software options, user popup menus, commands, processes, data flow, interface, and much, much more. What might be the cost in a breeder’s salary for say a year spread among several employees? The cost is real – put that down as your first figure. Call it: Software Design Specifications (SDS) – it is a big cost factor.
2. Actual Costs
Second, what will it cost to actually develop the system? Will you do it ‘in-house’ or contract out? In either case, you have to pay the salary for the staff in your company, or the contractors. If in-house, what if the staff leave? Replacing the IT staff with new employees who must spend months studying the code is expensive and costly. So what might be the salary costs, or the cost of the software company to write the system for you? Unless you are building a glorified breeder utility with limited possibilities, this will be hundreds of thousands of dollars to start – and that is really minimal. We know of a few government organizations that went to a local software company that had no real idea of the complexity involved. They were all too eager to develop a system (read “more business”!), said they had the knowledge, and provided an ‘attractive quote’. It was not a good experience in the end – they couldn’t deliver. Call this software development quote “SDQ” as a cost factor.
3. Lost Opportunity
Third, what is the opportunity cost during development? The development team will probably underestimate the complexity involved for a really solid, robust system. How long will it take? What if you are quoted two years, and after two years, it seems it will require six more months, then another year? Most companies are reluctant to ‘cut their losses’ at that point and license a software system. Management is loathe to admit a mistake, and the contractor will not want to lose the business. “Politics” seems to describe it at this point. Maybe you have heard of the old saying “don’t throw good money after bad”.
Meanwhile, the breeders will have this ‘sinking feeling’ that this is not going well. So how long will it take? The longer you wait, the longer your company loses the value of more efficient and powerful software and the ability to identify the best hybrid in your plant breeding pipeline. Your competitors are still “at your heels”. This is an intangible opportunity cost, but that doesn’t make it less real, it must be considered. By one company’s estimate I talked to, it can easily be a million dollars per year. You know your crops, breeding pipeline and market. Consider the opportunity cost as “OC”.
4. System Testing Time
Fourth, what will it cost you in time to test the new software? Testing has to be rigorous, and it takes time. You don’t want some critical bugs to surface at the worst possible moment when the software is fully in use. Remember Murphy’s Laws? If you “sign off” too soon with an outside software company, you can be sure it will cost a lot more to fix a bug after deployment. So, maybe a total of four month’s salary? Consider this testing cost as “TC”.
Can you pull it off?
Suppose you ‘go through all the hoops and pull it off’, develop a good system, in a reasonable timeframe. We assume that there was almost zero staff time in further design discussions during development (an unrealistic assumption) and the developers or your staff write the documentation and design a help system. We also assume the developers designed a solid, relational database architecture to build a powerful system to deliver maximum value for the dollars spent. That’s not easily discerned, but critical for the company. So the total cost so far is a minimum SDS+SQD+OC+TC. What might this be for you? Divide this by the cost of an annual license fee – how many years does this equate to?
Great, no annual software license fees! But software does not stand still; technology and user-needs constantly and consistently evolve.
What will be the cost of future modifications of the software, as well as upgrades as technology keeps changing?
If you out-source, will that company be there for you in the future?
What if they have a turnover in staff and in time there are few coders who know your software?
What will technical support cost you?
Who will do the software testing for future releases?
Who writes the updated user documentation?
As a simple minimal benchmark, what is the salary of one IT person in your company? That alone might be more than a software license fee. Two staff might be more realistic. The software does not run on ‘autopilot’. Anyone who flies a plane like that will ultimately crash.
What if the entire system needs to be re-written in say ten years because the software technology has changed significantly? Some of our new clients came to us precisely because the system they use “is getting old” and the cost to re-write was prohibitive.
So the foregoing costs might equate to a large portion of an annual software license fee, even though you own your own system. Not a comforting thought for “bean counters”.
I fully expect that licensing commercial plant breeding software will seem far more attractive and present good value. But there is “added value still”! Not only do you receive support, upgrades, bug fixes, and future releases with new technology as part of the annual license, but you will enjoy the input from breeders in other companies. I have seen in our training courses where breeders are amazed at a feature in our software, sort of like ‘why did I never think of that’? Cross-fertilization of software ideas – best realized from a plant breeding software development company with a proven long-term track record in the industry.
So before you embark on developing your own breeding software, count all the costs. You can be sure that not doing so will cost you. Focus on what your company does best – developing superior hybrids and varieties – and avoid distractions.
Dieter Mulitze, PhD. is CEO and Founder of Agronomix Software, which has been helping hundreds of plant breeders and variety testers manage and analyze their data, improving crops and creating better, higher royalty-paying varieties and hybrids.