I was lucky to learn this lesson early in my career. Many times, an organization will look for consolidation opportunities. Sometimes it makes a lot of sense to consolidate systems: there are cost savings involved, unit costs go down, you decrease the effort needed for change management…it’s the classic argument to create scale. A transaction is a transaction, right? So when is it a problem?
I was asked to assess two transaction processing systems early in my career. Business A was lower volume, large dollar transaction amounts….in the millions. Business B was much more volume, much lower dollar transaction amounts. Business A was eager for Business B’s volumes….why? The unit cost allocation would be much lower with Business B’s volumes in the denominator and the business P&L would improve. Business B was likely to be the surviving system, so they were for it as well. My recommendation?
No. Here’s why:
- Pipeline capacity: The pipeline has to be able to process every transaction by a certain time. If you haven’t built a pipeline for the maximum amount of throughput at peak times, transactions will fail. This is the one time you do have to “build the church for Easter Sunday” . Even if you get this right…
- Controls: there are always controls and queues for review…and they tend to need manual intervention: if they aren’t structured to handle large peaks, they will stop the flow.
- Who makes the call to prioritize when there’s contention? You’ll have to prioritize which transactions go…but who decides? Try finding a senior person late in the day to make that call.
- The impacted customer: often it makes sense to sacrifice the small dollar transactions. Except when it’s a little old lady who didn’t get her payment made and just wrote a letter to your CEO.
- Change just got more complex: now you have two completely different constituencies fighting for the same dollars to get projects done. Two completely different business with different needs. What data is needed for what group, for which regulator? Which need is more important?
Was my analysis unpopular? You bet. Was anyone willing to override it? Not so much.
Before you decide to “cross the streams” (see the original Ghostbusters), make sure you know when a transaction is actually not just a transaction.