Advanced Profile Synthesis
What is the difference between “traditional merge logic” and Advanced Profile Synthesis?
With the process known as “traditional merge logic,” only an exact match of the first name, last name, or email fields in two or more guest profiles will result in those profiles to merge.
Revinate’s Advanced Profile Synthesis, on the other hand, does not rely on an exact match of these or any other fields. It examines and compares the majority of the fields contained within a profile, and is sophisticated enough to overcome the following issues:
- Masked OTA email addresses
- Name changes
- Misaligned fields (the right text in the wrong field – i.e. Last Name text in the First Name field)
- Faulty text (abnormal text that is not applicable to its field – i.e. the word “Blank” in a First Name field)
How accurate is Revinate’s Advanced Profile Synthesis?
Revinate’s algorithm is highly effective and accurate. With that said, it has been trained to err on the side of being conservative. If the system is not sure, the default action is to not merge profiles. We care more about making sure we’re merging the correct information, rather than merging for the sake of saying that we can. The upside is that our algorithm is learning all of the time and, therefore, is constantly improving. The more data it has, the more accurate and certain it will become.
What is the benefit of using machine learning to merge profiles?
By using machine learning, you no longer need to get your hands dirty to clean your data, which is a very tedious, manual process. With Revinate’s Advanced Profile Synthesis, our algorithm will do it for you, thus saving you hundreds of hours that can be better spent generating revenue.
Why is Revinate’s Advanced Profile Synthesis important?
Revinate’s Advanced Profile Synthesis gives you one true view of your guests, and you’ll find that you have more loyal guests than you may have realized. As our algorithm cleans, deduplicates, synthesizes, and merges the raw profiles that we ingest from your PMS to form Rich Guest Profiles, the number of average stays at your hotel, or across your portfolio, should increase.
How often does Revinate merge profiles?
We run a property’s or group’s entire dataset through our Advanced Profile Synthesis automatically once every 48 hours. However, our existing “exact match logic” for the first name, last name, and email will continue to apply in real-time.
- Exact Match Example: If Caroline Brettell makes a new booking with the email address email@example.com and you already have her in your database as Caroline Brettell (with the same email), we will merge the profiles in real-time using our “exact match logic.”
- Advanced Profile Synthesis Example: If Caro makes a booking under Caro Brettell and you already have her as Caroline Brettell, the Caro and Caroline profiles will not merge until we run our daily Advanced Profile Synthesis merge (once every 48 hours).
Which fields does Revinate use to merge profiles?
We use a vast array of fields to ensure accuracy before merging profiles. The following are just a few that we take into consideration:
- First name
- Last name
- Phone number
Other techniques we use to help the algorithm identify a merge are:
- Possible phonetic matches for names (Stewart and Stuart)
- Possible nickname matches for names (Nick and Nicholas)
- Frequency of values in a database
- If a name is unique and thus less common, it is more likely to be the same person
- If a phone number, email, or address is too common, it might be junk
- Distinguishing between home and work addresses
Note: Transactional emails (confirmation, modification, and reservation), and most stay related emails (Pre-Arrival, Welcome, and We Miss You) will always go to the email address listed on the reservation, whether or not that email address is the primary email on the guest’s Rich Guest Profile. The same logic applies to double opt-in. OTA win-back emails are considered promotional emails; therefore, they will be sent to the primary email address in the guest’s Rich Guest Profile.