Individual Smart Campaigns vs Combined Smart Campaigns with Combined Filters

Anonymous
Not applicable

Individual Smart Campaigns vs Combined Smart Campaigns with Combined Filters

We are implementing a new scoring approach in Marketo.  We have 5 criteria we can score against, each with two choices.  That leaves us with 32 possible combinations of criteria and choices.  Each of those are ranked as either an A, B,C, D or E lead. Lead processing time is the main concern, as we upload a file with up to 30,000 leads in the morning, and need them to be processed as quickly as possible.

Which is best - 32 different smart campaigns, each with a possible combination, or building 5 different smart campaigns, and combining the different scoring combinations along with the logic in the filters?

When combining, we run into the issue of having up to 39 filters with filter logic such as ((1 or 2) and 3 and 4) and ((5 and 6 and 7 and ((8 and 9) or (10 and 11))) or (12 and 13 and 14 and ((15 and 16) or (17 and 18))) or (19 and 20 and 21 and ((22 and 23) or (24 and 25))) or (26 and 27 and 28 and ((29 and 30) or (31 and 32))) or (33 and 34 and 35 and ((36 and 37) or (38 and 39)))).

Are there potential issues with having a high number of filters in the smart campaign? Which would process leads faster, 32 campaigns running concurrently, or 5 campaigns with lots of filters? Which is the safer approach?
Tags (1)
4 REPLIES 4
Josh_Hill13
Level 10 - Champion Alumni

Re: Individual Smart Campaigns vs Combined Smart Campaigns with Combined Filters

Batch campaigns would be safer.

I don't completely understand your approach, but that's not impt. Do you always upload 30k records?? If it's once, then it doesn't matter so much. Every day, yes.
  • 32 triggers will add to your system's load. They will not all fire at once and they could create race conditions if your smart lists are dependent on each other. This may not be as bad as I make it, so I would tend to go with this one.
  • 5 Campaigns with 39 filters will also be a giant load and probably break. Don't do this.
So my recommendation is to choose Option 3:
  • 01 - Scoring Entry - this campaign brings in everyone who hasn't been scored. Then it requests one of 32 campaigns based on the criteria involved. I think you could even get this down to choosing one of the 5 campaigns, then letting those filter further. Hard to say without seeing this.
Anonymous
Not applicable

Re: Individual Smart Campaigns vs Combined Smart Campaigns with Combined Filters

Thanks for your thoughts.  Here's more detail that might be useful.  We upload our lists every weekday. All campaigns are triggered, and need to be. The scoring smart campaigns are not dependent on each other.

The chain of events is as follows:
  1. Leads are created or updated, and this triggers partition assignment.
  2. At the same time, we are also triggering "agent assignment" smart campaigns based off the lead being created or updated, and the flow steps "assign" the lead to an agent or to the general pool
  3. Once the agent assignment happens, emails are then triggered.  The agent assignment needs to happen first because the emails are customized using the agent contact information.


Emails need to go out as the agent is assigned.  The processing time of the file upload varies each day depending on the size of the file.  On Mondays, for example, we can have up to 30,000 leads, and processing can run through 4pm in the afternoon.  Emails need to go out starting out in the morning, as it drives calls to our agents or the call center and waiting until 4pm means a call load starting at 4pm, and lower email open rates.

What we are trying to accomplish is inserting the scoring piece before the agent assignment so that we can assign based on the score.  Higher scoring leads are routed to an agent, while lower scoring leads are routed to a pool of contacts directing leads to contact our call center.

So it would look like:  Scoring -->  Agent Assignment --> Email Send

To give you an idea of what our processing looks like, at any given moment after the file upload happens, we can have up to 167 campaigns running in the Campaign Queue - including email sends, agent assignment and partition assignment.

Here is the scoring criteria.  We score based on whether there is a discount attached to the lead or not, lead time equates to how many days out they want their delivery - over or under 90 days out from today's date, mileage over or under 500 miles, whether the origination locations for their delivery are categorized as good or neutral locations, and whether the destination locations for their delivery are categorized as good or bad locations.

0EM50000000TA2J.png
Anonymous
Not applicable

Re: Individual Smart Campaigns vs Combined Smart Campaigns with Combined Filters

Hi, 

I think you can do this with just 1 campaign and some tokens.  However, you will need to change the way you are thinking about scoring.  

What if your 'Score' is based on a total of 100 points and you attribute a weight to each value? Looking at your grid it seems something like this could work: 

Total Score: 100 pts
Destination 40%
Lead Time  25%
Discount     20%
Mileage      10%
Origination  5%   

So if destination is 'Good' add +40 to score but if Destination is 'Bad' than 0 points are added.
If Lead Time is your ideal value, add +25 to score but if Lead Time is a negative value than 0 points are added. 

This model could work but the proposed scoring model may push the lowest scoring D leads into the E bucket and the highest scoring D leads into the E bucket.  Play around with some calculations and see what happens, hopefully it is accurate for how you are wanting to segment. 

If the above checks out for your setup than you can use ranges and attribute a letter appropriately. 
95-100 = A
80-90 = B
60-75 = C
35-60 = D
0-30 = E

Just another option to consider 

 
Anonymous
Not applicable

Re: Individual Smart Campaigns vs Combined Smart Campaigns with Combined Filters

We were using weights and tokens before but we were still getting the neutral to bad Origination to Destination combinations in the wrong scoring group.  Your suggestion is good and I'll play around with the scoring to see if we can it where we need it.  Thanks!