Batch Scoring Every Behavioral Action?

[Anonymous]

My scenario is as follows:

 

1. Lead has visited 5 web pages so has a score of +50.
2. We reset all scores due to having updated our lead scoring values.
3. We run a batch scoring campaign for web visits.
4. New score is now only +10, as the batch campaign only applied a score for one web visit before moving on to scoring another lead.

Is anyone aware of a way to retrospectively batch score on every behavioral action a lead takes, or is this a limitation in the current Marketo version?

Thanks in advance for any replies!

[Kimi_Heskett1]

I think you can only batch that behavior once in that fashion. You could build smart lists and say has visit 1 page, run flow + 10 pts., has visited 2 pages, run flow another +10, etc etc etc. 

[Anonymous · Accepted solution]

Hey Adam:

Good question. You can do it, but the question is more of how much time do you want to spend doing it versus how exact do you want the data, because this can be a very manual process. I think the answer is going to be having lots of smart campaigns with filters, some math logic, and some slick requestable campaigns.

The question is how many steps and how granular you want it to be. If you look at the metrics and people are visiting your website every single day for 90 days, then you could be looking at 90 separate smart campaigns doing exactly this. OR if you see that 90% of leads visit 20 times, and there's a long-tail of the rest going up to 200 visits, you can do something like below.

Here's the formula:
"Visited web page (is any)" & "Min number of times = x" -> Increase score by (value of [this number of times minus the number lower below it] visits)

Here are example smart campaigns:

• Filter "Visited web page (is any)" & "Min number of times = 20" -> Increase score by (value of FIVE visits)
• Filter "Visited web page (is any)" & "Min number of times = 15" -> Increase score by (value of FIVE visits)
• Filter "Visited web page (is any)" & "Min number of times = 10" -> Increase score by (value of FIVE visits)
• Filter "Visited web page (is any)" & "Min number of times = 5" -> Increase score by (value of TWO visits)
• Filter "Visited web page (is any)" & "Min number of times = 3" -> Increase score by (value of ONE visit)
• Filter "Visited web page (is any)" & "Min number of times = 2" -> Increase score by (value of ONE visit)
• Filter "Visited web page (is any)" & "Min number of times = 1" -> Increase score by (value of ONE visit)

So let's say a person visited 10 times in the last 90 days (because 90 days is all you get with storing web activity history!).

1st visit = +1
2nd visit = +1
3rd visit = +1
5th visit = +2
10th visit = +5
Total visits = +10

Excellent! It works!


Now to make it scalable, if you ever plan on doing this again, or if you ever want to tweak the scoring. Which will be the case, even when you don't think it will be at the moment. Might as well do it right and make it replicable the first time.

• Make all of these campaigns (which would be batch campaigns) requestable by a Marketo flow action. Create a general batch campaign that requests all of these mini-scoring campaigns. This makes it so you just have to run that single batch campaign once to go through all of those individual page scoring campaigns.
• Tokenize those scoring values. If you're doing a more-than-one score (e.g., jumping from the 5th visit to the 10th visit) to save setup time, add multiple separate flow steps with each one increasing the score by a token, not a hardcoded number of the final product.

The idea would be to build up a nexus of requestable scoring campaigns that can be triggered by running a single batch campaign. Of course, you can do this same incremental logic with more than web activities, as you should. If you ever chose to redo behavior scoring, you can simply change the token scoring values, run the master batch rescore smart campaign, and all behavior/lead scores within the last 90 days will be recalculated.


Best,
Edward Unthank
Marketing Operations Specialist
Yesler

Also, I used this discussion as an excuse to write a blog post outlining how to set up a robust behavior rescoring program, which you can read here!

[Anonymous]

If you are keeping the same behavior scoring model, the more elegant solution is to keep the behavior score but reset the demographic score.

Typical process:

1. Create new score fields for behavior score and demographic score
2. Score based on the old demographic scoring model in the field "demographic score"
3. Subtract the demographic score from the total score, and move this value to the behavior score
4. Rescore demographics based on the new model
5. Combine the two scores into the total score

[Anonymous]

Thanks to all of you for your replies on this.

Edward: That's a fantastic answer which I'm sure will benefit many community members in the future, myself included. The scalability tips are also good productivity boosters which anyone who manages lead scoring could (and probably should) utilise. Thanks once again!