Data migrations can be challenging initiatives at the best of times, but when they start to go wrong they can be devilishly difficult to turn around and keep afloat. In this article I share three simple ideas that may help you gain control of the project and bring that go-live date back on track.

Tip #1: Gain an external perspective

A lot of data migrations involve multiple parties, and this often creates a "them-and-us" situation where relationships break down.

For example, the client organisation may be frustrated at having to improve the quality of their legacy data. Meanwhile, the supplier grows weary of having to wait for the client to prepare the data for migration before testing can begin.

Situations like this can quickly result in broken trust which is why it’s very useful to have an independent data migration expert help take a holistic view of the project. The client often lacks extensive in-house data migration experience so getting an independent set of eyes to assess progress and provide recommendations is ideal if the project heads south.

Tip #2: Reassess and re-plan

One of the big mistakes data migration project leaders make is to soldier on into the storm, hoping they’ll get a break and turn the project around. The reality is that when deadlines get squeezed, best practice techniques get ignored. The end result will be disappointed system sponsors as the target implementation either overruns horribly or is riddled with poor-quality data.

Ignore the original plan; it was only an estimate anyway. Now you’re faced with the reality that you need to take stock and revisit those original forecasts. You should have a lot more data points to help you determine how long your team takes to complete tasks such as data quality management, migration mapping and testing. Use this information to re-evaluate your estimates.

A big mistake is to think of data migration projects as some linear production line with contiguous jobs rolling along the assembly belt. Data migrations are actually cyclical in nature and follow a strategy of: assess, plan, design, revise, design, build, revise, test, revise, go live.

It's impossible to create a perfect data migration plan from the outset. You just don’t have that degree of knowledge, even after a pre-migration impact assessment. Knowledge has to be gained during the project, which requires regular reassessment and re-planning. Project leaders need to accept that they're not building a piece of software with a classic waterfall methodology. You’re building a house on shifting sand; you need to improvise, adapt and overcome.

In particular, you need to perform a detailed 360-degree review of where the project is delivering or under-delivering. This should be fairly obvious, but this data migration checklist may also prove useful.

Tip #3: Tighten up change control (particularly data quality)

When deadlines start to drift on your data migration, non-existent change control can be the first symptom of a panic-stricken team as they strive to get over the line. Team members start to cut corners, resulting in the project leader losing visibility over the state of the project.

The main lifeline into the project for leaders is through regular meetings, progress reports and other project artifacts. Accurate change control allows you to observe which tasks are completed or awaiting further work.

For example, the data quality rules process requires staff to "check out" data subjects and work on them until fully signed-off and ready for migration. By keeping your change-control process well governed, you can prevent any unwanted surprises at run-time. As a project leader, you need to be proactive. Make sure change control is being applied correctly by observing the team in action and listening to their concerns.

What other tips do you recommend for turning around a failing data migration? Welcome your views.