Algorithm behind duplicate check rule

[Former Member]

Our customer is asking us what the algorithm is behind the configuration of the duplicate check rule.

For example let's assume I want a duplicate check based only on the customer name. What is the effect of moving the weighting factor up and down. Is 100 used for exact matches and does the check become more fuzzy when going to 0?

What happens when you mix up several parameters (e.g. name and address)?

How are name and address defines? concatenation of specific fields? Which ones?

It would be great if SAP could document this Black Box. The alternative for our customer is to waste hours via trials and errors....

[Vinod0311 · Accepted solution]

Hi Thomas,

That was an awesome response, was always a grey area.

Meanwhile, for the below scenario, would like to understand if the algorithm still applicable.

Name = Bond
E-mail = j.bond@mi6.com

Let’s say there is already this DUPLICATE contact existing on the database:
Name = Bond
E-mail = j.bond@mi6.com

Same customer name and email address --> This is more applicable in Email/SMS to Ticket.(Ticket creation automatically linking the customer - when the customer send an sms or email)

Wished to understand what logic the system applies, to pick the customer when there are duplicate records.

Gab said the team is looking into, but this is quite an urgency for our implementation.

Is the Levenshtein algorithm still applicable in this scenario ?

Many thanks,

Vinod

[former_member541649 · Accepted solution]

Hi Vinod,

The email/sms design doesn't expect duplicate records and considers email address to be a unique identifier of a customer, hence there is no duplicate check incorporated there.

If there are multiple records found, the system picks one of the records present in the database. It usually picks the oldest active customer record, but it is highly recommended to clean up the master data to avoid such situations altogether.

Hope this helps.

Regards,

Parul

[Vinod0311 · Accepted solution]

Many Thanks Parul,

Makes sense to clean up the MDM, but there is absolutely less time available, for the agent to respond to the customer and identity the booking, coz the expected is the Ticket should link the right customer who has the recent Transaction/Booking.

I also performed check if the SMS/EMAIL to case is picking up the oldest or recent customer record, but the result is inconsistent

[Former Member · Accepted solution]

Hi Vinod,

We will review the logic used to pick the customer in case many records have the same email or phone number, and make sure it behaves consistently.  But the best we can offer is something similar to "always pick the oldest active record", not anything related to Tickets or Bookings.

This may require some enhancement on our side, so we'll add it to the backlog for the next release.

Thanks,

Gab

[Vinod0311 · Accepted solution]

Gab - Many thanks your support, as always you are fantastic..!

Cheers,

Vinod