What I learned from merging SFDC Orgs

November 09, 2017

I recently took on a project to merge a foreign separate instance of Salesforce into our main instance of Salesforce. The timeline was extremely tight and short, with very little wiggle room - I am talking about 4 weeks from when the instance would be shutting down for good.

The first step of getting our heads around this project was coming up with a project plan. I used Confluence to create a page that contained all the different sections with checklists of all the pieces of information we would need to collect. I must say, we got super lucky in this case. Despite the instance being in a foreign non-latin based language, their org structure was super simple, which made the collection of information much easier. (If you were wondering, they were in Professional with no API access so you can imagine the lack of things you would need to worry about in that case).

Here is the list of items I found to be important to collect:

Once we had the plan together and a checklist to keep me somewhat sane, we began the task of going through it all. Working with both our internal teams and the team abroad, we collected the answers, which took about a week. During this time, I was also prepping all of the foundational groundwork in a dev environment. This included creating the new fields, setting up the new roles, creating new page layouts, etc. Our plan was to give ourselves a week+ to get it all together and replicated in our full sandbox, leaving us with about 4 days to have users test and then roll out to production. This gave us an extra week in case something went horribly wrong.

While all intentions are a best-laid plan… or however that saying goes, our plans for testing the merged changes did not quite go as hoped. We had difficulty getting users correctly set up in time for the testing as we didn’t have the correct emails for everyone and therefore only about 75% of the items got tested before we went live. Luckily, we knew nothing could break in any catastrophic way (no APEX thank goodness) so we continued on as planned regardless of testing not getting to 100%.

Quick comment on getting the users from the old org to the new one - this was the most difficult step in the entire process. Between 2 different email domains and some of the users already existing in the destination org, it became a tangle of excel sheets and reports working to get it all lined up correctly.

Once testing was done, we needed to push first the foundational work into production and then the data. Given I had already moved all of this once from dev to full, this part was a walk in the park. The data piece took about 2 days to do (and that’s for a relatively small amount of data) - we did need to turn off validation rules, workflow rules, and triggers in order to easily get the data in there.

The last step in this process was the post-deployment step. This has been one of the more lengthy pieces of the process. As with any big project, there is always a bit of cleanup and polishing to do at the end and this merge left us with several days of that. I think, honestly, that no amount of additional testing or pre-work could have been done to avoid it. So, if you are going through this process, don’t beat yourself up about it; just plan the time in advance to take care of it. The cleanup mostly encompassed sharing rule fixes, role hierarchy changes, and dashboard and report updates.

My 20/20 hindsight takeaways of this whole project are: