Can someone please advise whether it is faster to update the same value in one data field for a large quantity of leads (100K+) by:
a) Smart Campaign
b) List Import
I know that is a tricky question and I'm adding the best experts known in Marketo world to help us out?
We have seen that marketo takes 7 hours for updating 1 field with same value for 170K persons through a smart campaign "data value change" command.
Quite surprising in a database update world, we don't know if it's due to GDPR's effect that the marketo server is so slow (maybe everybody in marketo is making update right now).
We wonder if there is way to perform this kind of action faster, maybe through a excel import?
We are talking about 24k records /hours which is not that bad. First look at you campaign queue to see whether some other, higher priority activity is slowing down this one. If you are running a smart campaign which first flow step is a data value change, it makes it a medium priority one and that alone could explain the duration of the process.
Thanks for your reply.
Is there any documentation I can refer to or it's based on your knowledge?
I'm looking a marketo documentation about this limitations and how priority are defined in order to prepare my fix.
Thanks for your help.
we don't know if it's due to GDPR's effect that the marketo server is so slow (maybe everybody in marketo is making update right now)
Ah, don't think it's GDPR.
But there is a limited rate of intake for any of the import processes, not only because of some overall resource governor, but also because (we can guess from the behavior) it's not one big transaction but rather row-by-row. (Otherwise, you couldn't get a row-level exception list from the bulk import endpoint, for example.)
If it were truly a SQL-level bulk import with no resource caps, then naturally it would be many, many times faster if run by itself -- but also would be far less cooperative with concurrent processes.
My personal preference is to do this via bulk import or list, though I don't know if there's a substantive difference.
This is from my personal experience. I like list import as its faster to make mass updates and doesn't queue up my campaigns and cause a bottleneck. The setback with this that I've noticed is that I have a program if an email address changes, update invalid email to false if it was true. If you have something similar, turn it off temporarily so it doesn't cause invalid email addresses to revive by accident just because of a capital letter difference. kdelgado vs Kdelgado.
One other thing is that list imports will only grab one "person" that matches the email address. If you have duplicates, it may not update the correct one you wanted. You may want to do the list import and then run a smart campaign to cover anything that may have been missed(field is blank or not [value]), which should be relatively low. Or export and cross-check, I do this sometimes with excel so I can verify.
Also when doing list imports, its not suppose to create duplicates with "persons" with the same email address but I see it happen still(very few though). I use lead source on my side that doesn't get overwritten. I make lead source, "LI_persona-update_20180528". If anything has that value and was created in the past 2 days, just to be safe, I review and merge/remove if necessary.
Another also to note, if you're updating a field that is synced to your CRM, you may want to turn off the sync temporarily so it doesn't use up too many api calls. Once the update is complete, then go ahead and turn on again. I prefer it this way as it bulk updates instead of one at a time or a few at a time. It uses less api calls from what I've seen. Hope the above helps!
For what it's worth, I tried updating one field on 600k+ records with a smart campaign. I killed the campaign when it had not completed after 4 days and we did a list import which was done the same day.
For simple, large-volume changes, Bulk Import is your best bet - it works in 1k chunks and does direct writes to the database. Smart Campaigns are overkill for that amount of change. Performance can vary based on a few factors, mainly standard attributes versus custom attributes.
I actually talked about both in my Under the Hood session at Summit. And by total coincidence, I just recorded the audio to the slides yesterday. Enjoy!