Confessions of a Recovering Data Analyst - Why Software Migrations Fail!
“After hundreds of thousands of dollars, 9 months of implementation, and numerous hours from our staff – we STILL have reports that are inadequate and fragmented.”
-said most executives more than once…
Did you know the overwhelming majority of new software implementations fail? The odds are you’ve probably suffered through a failed implementation yourself or known someone who has. The fallout caused by the loss of the investment, missed expectations with stakeholders, and the failure to achieve the promised result can be a nightmare. A major culprit is data migration - the critical step where legacy data is gathered, joined, and uploaded into the new tool. On the surface, this doesn’t seem unduly complicated but ask anyone who’s done it and they’ll tell you it is a very challenging process that precious few software vendors are equipped to handle. Here are just a few examples of what vendors must be able to do to succeed:
Surgically combine overlapping data from multiple sources.
Asset records usually exist in multiple legacy systems (e.g. CMMS, ITSM, DCIM, Sharepoint) with a degree of overlap/underlap, but only rarely are there clear-cut identifiers to match assets between systems. As a result, vendors use their best judgment and often unknowingly upload duplicate assets into the new tool, which skews reporting on items such as capital replacement, maintenance planning, failure rates of particular makes/models, etc.
Solve for ambiguously duplicative values
“ACME & Co. Inc.”, “Acm”, and “A.C.I.” are probably just different representations of the same manufacturer... or are they? If they aren’t, then merging them will corrupt your data. If they are, then leaving them separate will skew reporting. Without intimate knowledge of thousands of facility equipment manufacturers, most software vendors are faced with a lose-lose situation.
Efficient field collection for missing values
Some missing information requires input from knowledgeable site staff (e.g. serial numbers, install dates, capacities, or clarifying if a UPS is ‘static’ or ‘rotary’), but in many cases the answer to one question is needed to inform the next question, which for most vendors necessitates a repetitive and time-consuming cycle of preparing, distributing, collecting and merging spreadsheets. Before long, this ‘second job’ wears thin on the site teams, and often the process is abandoned before it’s complete, resulting in gaps in your data and analytics.
With MCIM by Fulcrum Collaborations, we use an innovative inference technique to address overlap/underlap and triangulate around ambiguity while standardizing and enriching legacy data en masse. We also provide an easy to use interface for targeted field user input, which learns characteristics of existing data to nudge users towards standardization when they create new data. The net result is clean and complete data, with far less work for site teams, and an extremely rapid implementation timeframe of only 30-90 days!
If you’d like to see first-hand how we solve these common issues in data migration, request a demo and we'll show how MCIM can make your data work for you, not against you.
Originally published Mar 26, 2018