A Question of Granularity
There is a long simmering issue coming back to the front burner these days. It’s the question of best of breed software vs. a single system. I’ve been giving it a lot of thought and realized something.
The old discussion says that best of breed opens up application areas to greater competition from more vendors. This competition drives normalization so that everyone can build to the same open standards rather than proprietary architectures. This approach worked for the relational database and SQL, PCs and servers, and standardized programming languages just to name a few things.
Alternatively, those supporting the solo idea say that for complex processing having a single throat to choke is a valuable asset. Who is right? Can the answer depend on a tiebreaker of sorts? I think the answer is beyond this question and ultimately comes down to a question of granularity.
In the software business we’ve seen the industry veer from one extreme to another. Early in a lifecycle, it seems, vendors merge and integrate systems to produce that single solution but it may be highly proprietary. Those proprietary systems are an emerging vendor’s best defense against a copy cat coming in and taking over.
Often best of breed solutions pop up when developers see opportunities to improve a process or even a sub-process to optimize it. The best of breed approach basically says that the monolithic solution can’t be great at everything and that customers deserve great. That’s true but the idea has a half-life because the longer a suite is in market the better it gets and at some point a critical mass of customers won’t even consider the alternative.
Today we see vendors like Oracle leading the charge for the single vendor idea saying that its products are engineered to work together. That would be the argument for the sole source. NetSuite argues from the same premise. But we also see companies like Salesforce with a massive ecosystem of partner applications that offer specialized apps that the company does not provide. Salesforce does deliver a very good development platform in Force.com and API that its partners use to develop their solutions. In this case I’d say that the Salesforce solutions involve such new processes that they are functioning like the early market vendor with a high walled garden while still offering aspects of best of breed.
This is a bit different from conventional best of breed in that the Salesforce partners more or less pre-integrate their solutions via the platform so that the only difference between a Salesforce application and a partner application is often whose fingers did the work. That’s why I would suggest that Salesforce’s approach is more like the single provider than the best of breed approach from just a few years ago.
So to me the question is not one of single vendor vs. best of breed. I think that’s a false dichotomy. Whether or not we realize it we’re all in a best of breed era and the only question is at what level of granularity? I don’t know anyone who seriously thinks that some level of best of breed is NOT a requirement today — there are simply too many demands and options to expect a single provider to do it all unless all the software companies of the planet merge.
The best of the best of breed solutions will arrive at an appropriate level of granularity that optimizes internal lines of communication within the system while incorporating external best of breed solutions at the periphery.
That dichotomy will be different from system to system. For example, CRM has done a good job of integrating several generations of applications including whole systems like call center and social media as well as specialized hardware like IVR gear to produce good solutions.
ERP seems to be different because the back office is a bigger thing that needs to coordinate many more moving parts. Frankly, I think the critical mass of application solutions is just bigger in the back office than in the front office. So the discussion of best of breed has to be qualified by which part of the business we’re referencing.
In ERP I don’t think you gain anything if you suddenly offer best of breed GL and AR distinct from accounts payable or some of the manufacturing systems like supply chain, product lifecycle management and similar things that need to be tightly integrated. On the other hand, HR was never that tightly integrated with the back office and it was more of a traditional offering that went with the back office because it paid people and the back office was where the money is. But these days with a proliferation of human capital management systems, training, hiring and things like them, the ties are less strong which has opened HR up to best of breed offerings for these newer functions.
Billing and payments is another area that has recently come up for best of breed renovation. When those functions were associated exclusively with manufacturing it all made sense. But today the proliferation of the subscription model has placed new demands on back office billing that it was not designed to handle. Subscription billing and payments has become a satellite of conventional ERP and truth be told companies like Zuora, Aria and all the others in this niche, do a better job of managing subscriptions than old style ERP can. So, again, we are seeing an area open up to best of breed approaches.
For me, you need to ask about the level of granularity at which you are viewing the business problem in order to determine the answer to the bigger question. Then, too, we haven’t even looked at the new business processes that are being glued onto the front office through social techniques. It seems like the majority of new business processes are going to the front office and the back office is largely settled business. Even subscription billing is looking more like a branch of customer service than part of ERP. So, the issue isn’t best of breed vs. an integrated solution, it’s more about how much best of breed can you handle before you have too many balls in the air. I think the answer is it depends.