Thought I'd share a very recent mishap in our Marketo instance to bring a smile to your faces.
We've been re-building our engagement programs, with 22 pre-existing content programs that needed to be be re-built as new types for reporting purposes. We wanted to make sure the members of the old programs were transferred to the new programs so that we could maintain the integrity of the overall program and not send out the same content again to the same leads.
Because we were introducing a new type of program we spent a lot of time getting the new emails, new tokens and new campaigns right in one before then cloning it out. In the campaigns are a set of of batches to transfer members in from the old campaign. There are 5 campaigns per program, one per status we're moving, so 110 overall. We couldn't see a way to create one with 5 choices as there seems to be no choice for membership of a different program. On reflection we probably should have pushed from the old campaigns, rather than pulled in from the from campaigns. Anyway - another lesson learned there.
Clearly, sitting down and firing 110 campaigns did not appeal to me, so I realised a single request campaign could call all 110. Of course I had forgotten to do this in building them so I sat down and edited each campaign to trigger on campaign requested. Each transfer in program was already filtered for members of the appropriate old program.
Then I built a control campaign which filtered on members of the 22 programs that need to be transferred, with 110 flow steps of "Request Campaign". Since I knew each campaign was filtered already looking for membership of the previous program I did not use Choices for "if member of program".
I then happily sent through the 20,000 leads that needed to move programs. The next day when I checked the results I saw that the campaign was still running - so far with around 600,000 results. How could this be? I only had 20,000 members to move. It soon dawned on me that although the requested campaigns were filtered to fire only when the lead was a member of that program, the actual main campaign was requesting each of the 110 flow steps for all 20,000 members. This meant had created a campaign that would fire 2.1 million times in total before finishing.
I worked out that it took me about 2 hours to edit and build the overall request campaign, which we then calculated was going to take over 48 hours to run. My colleague(@David da Silva) , once he'd stopped falling off his chair laughing, aborted the campaign and ran each of the 110 individually as standard batch campaigns in a lot less time than the two hours I originally spend to save time!
Request Campaign and automation in general can save a lot of time, but make sure you properly think it through first!