Best of Breed or Best of Need
If your business is all about a particular unique selling point, then build that specific to your needs. However, ask yourself, do you need a unique mechanism to add a product to a basket? Do you need a specific way to capture a credit card payment?
Odds are that the answer is no to many questions in an enterprise domain.
Sometimes you will see that MACH architectures laud the ability to compose all of the best of breed solutions together. This is indeed an advantage, but to plumb all these disparate software together requires a lot of investment in the glue to integrate them.
Operate with a mantra of “build only what’s unique to you, buy everything else”. This is good in the majority of cases, however, I’d go one step further in buying pre-composed solutions to macro domains of functionality where there is no need for your enterprise to differentiate.
For example, can you operate successfully with order management, fulfilment basket management and logistics all done in a generic manner (configured for your business, of course) and then have a very customised and unique content approach.
Do you have millions of products to index and filter and require algorithmic or complex AI based recommendations and conversational UIs? Or are your customers performing relatively simple text searches against a catalogues that is less than 100,000 SKUs?
You should also be asking yourself about how your data is being managed. Having data spread across many best of breed solutions means that your consistency of data will take a hit. The CAP theorem is something I apply to all my architectural analyses and it states:
Any distributed data store can only provide two of the following three guarantees:
Consistency
Every read receives the most recent write or an error.
Availability
Every request receives a (non-error) response, without the guarantee that it contains the most recent write.
Partition tolerance
The system continues to operate despite an arbitrary number of messages being dropped (or delayed) by the network between nodes.
from: https://en.wikipedia.org/wiki/CAP_theorem
So, as you add more best of breed solutions, your data becomes more and more volatile.
What about the operational overhead in managing your own software? Part of the benefits of microservices is great flexibility, but with many moving parts comes a large operational overhead. If you can reduce that surface area of operational management required, then you will find that your technical team can concentrate on functional enhancement, rather than keeping the lights on.
Operationally, many moving parts is a challenging landscape to play in, but at the same time, you’ll also find that the business user experience will be affected by having many so called ‘best of breed’ solutions in play in the enterprise’s content and data management ecosystems. Consider having one system that’s not quite the very best at a single domain, but offers a good enough user experience to manage customers, products, orders and promotions from the same user interface. Now consider the situation where you have a ‘best of breed’ solution managing each of those systems and you as a business user have to manage and maintain the data across all those systems. difficult.
Takeaway
Identify what makes your enterprise unique. Once you’ve pragmatically itemised those elements, find a vendor that does your key functionality well. If you’re content focused then base your main vendor decision around CMS providers. If you’re commerce focused, then base the decision around a commerce vendor.
Apply some basic heuristics to this search. Find a vendor that covers a wide gamut of your required capabilities, even if it does some of them not so well. Find a vendor that embraces composability so that when you need to build something unique for your business, the pain described above is minimised.
The MACH Alliance has attracted many of the best vendors in the Cloud Native and API driven space. Many of these vendors have excellent and even class leading credentials in a particular focus area, coupled with good enough capabilities in closely related areas. commercetools for example has excellent pure commerce capabilities, but average search capabilities. These vendors are good examples of what we are talking about here. Best of breed for domain elements that are close to your business heart, but not unique, and best of need for those elements of functionality that you need, but don’t need to be highly customised or unique.
Finally, find a team that will compose your new enterprise architecture by providing the glue required.
Find the ‘best of need’ for your business. Best of breed for those areas that are not core to your business will often be wasted effort for little return on the investment.
