Grasp Hotel Normalization is a tool that collapses the various address aliases for a hotel property into a single master hotel property to allow for more accurate hotel reporting in GraspDATA.
The problem with hotel information
Information for a single hotel property (name, address and phone number) can be input in various ways across different systems - even though it is for a singular physical location. Because of this, without cleaning and normalizing this data (marrying up the various “aliases” for a single address), it becomes nearly impossible to accurately analyze and report on bookings at individual hotel properties.
What is Grasp Hotel Normalization?
Grasp Hotel Normalization is a tool that collapses the various address aliases for a hotel property into a single master hotel property to allow for more accurate hotel reporting in GraspDATA.
How exactly does Grasp Hotel Normalization work?
Why should hotel data be normalized?
What are the benefits of normalized hotel data for Reporting?
What will I get with Grasp Hotel Normalization?
Would Grasp Hotel Normalization “fix” the various names for a hotel property?
Which data points are normalized?
Which data sources are normalized?
With what frequency is hotel information updated?
How long does it take to normalize my data?
How do you maintain new names, new properties etc.?
How does Grasp handle corrections?
How do the reports incorporate hotel normalization?
What data sources are normalized?
Is Grasp Hotel Normalization exclusively for properties in North America?
Do I need to purchase other Grasp products in addition to Grasp Hotel Normalization?
Can GraspDATA be utilized during the time hotels are being normalized?
What is the difference between Grasp Hotel Normalization data and the info in the back office?
Do corrections and normalizations apply to only customer (DK) or all customers?
How is Grasp Hotel Normalization installed?
Implementation timeline and hours for training?
Will Grasp Hotel Normalization make my reports run slower?
How do I interpret the QC report?
Will Grasp Hotel Normalization update my back office too?
How exactly does Grasp Hotel Normalization work?
- There are typically three pieces of information about a hotel – name, address and phone number. And their entry and accuracy entirely depends on the agent. Any or all of these could be accurate (or not), so all three are taken into account when determining the hotel’s location. Examples:
- Hotel name can be abbreviated, misspelled or missing words in the name.
- Address can be exact, of relative, such as “London” without a street address, or incorrect.
- Phone number can be missing digits, be completely wrong or non-existent.
- In order to resolve if these three pieces of information actually point to a hotel location, third-party mapping services plus internal master tables are referenced to derive if these three pieces of information truly correlate to a hotel property.
- First, an attempt is made to geocode the address to produce a latitude and longitude, to locate the property.
- If the raw address does not produce a lat/long, Grasp Hotel Normalization tries to auto-correct the address using outside address verification tools (which work across various countries). If it can be auto-corrected, the address is then geocoded.
- If auto-correcting the address does not return a result, Grasp Hotel Normalization scans the address to try to derive a city and/or country. The scan is based on city population, so if “London” is found in the address, the first assumption is London England (not London Ohio) because London England has a larger population, and presumably more hotels.
- If Grasp Hotel Normalization finds a city and/or country, it is geocoded, thus providing a center point (lat/long) around which to search for the hotel.
- Once the geocode has been determined, Grasp Hotel Normalization conducts a search around that geocode for the hotel name. This is necessary because to only geocode an address does not necessarily mean that address (lat/long) is actually a hotel building.
- If Grasp Hotel Normalization cannot find a hit on the raw hotel name, it starts cleaning the name by fixing spelling errors, expanding abbreviated words, adding keywords (“hotel”, “resort”, “spa”) etc., then searching by the cleansed name around the geocode.
- If Grasp Hotel Normalization cannot find a “hit” on the cleansed hotel name, within a reasonable distance from the geocode, it then scans the hotel name to derive the hotel chain. If the hotel chain name is derived, a search is conducted around that geocode.
- If searching by hotel name around the geocode does not work, Grasp Hotel Normalization then falls back on the hotel phone number to search a business database, to derive the address from there and see if it is within a reasonable distance from the original geocoded address.
- If Grasp Hotel Normalization cannot derive any geocode (lat/long) from either the address nor the phone number, it will search the entire world for the hotel name using the raw hotel name and cleansed hotel name (similar to checks in #4 above). If we have a “hit” on this, and it looks reasonable (it is a hotel and its address and/or name contains similar words to the raw data), Grasp Hotel Normalization assumes this is the hotel location.
- The process continues for each consecutive distinct hotel name/address/phone number.
- As distinct hotel name/address/phone number are geocoded (“normalized”), Grasp Hotel Normalization then collapses them into a single master hotel property record (MasterHotelPropertyID) which is used for enriching hotel vendor information for reporting. This allows for analysis to view for example “properties in San Diego” or “total sales for all Hilton properties in Mexico City”.
Why should hotel data be normalized?
Information for a single hotel property (its name, address and phone number) can be input in various ways across different systems even though it is the same physical location. Without cleansing and normalizing this data (marrying up the various “aliases” for a single address), it is nearly impossible to analyze and report on bookings at individual hotel properties. Questions such as “how many hotel bookings do I have in San Diego for conference X” cannot be answered without normalizing hotel addresses.
What are the benefits of normalized hotel data for Reporting?
- Accurate reporting of spend/sales, number of bookings, average nightly rate etc. at individual hotel properties and chains.
- Accurate location information to locate travelers (where are my travelers?) staying at hotels within disaster or high risk areas, or in case of emergency.
- Ability to distinguish between preferred or non-preferred hotels (data feed required - tbq).
What will I get with Grasp Hotel Normalization?
- Approximately 70 standard reports are hotel normalization-aware. This toggle appears on the Report Run, and Report Scheduling web pages:
- Additionally, up to 5 custom/standard reports (non-Report Builder) will be configured as part of the implementation to incorporate hotel normalization (the toggle feature above will be available).
- The following dashboard widgets are provided to monitor the normalization process.
-
- Line - Grasp Hotel Norm - Hotel Properties Normalized. This widget provides an overview of the normalization counts such as total hotels normalized, hotels normalized automatically via code (machine) and hotels normalized by a human.
- Pie - Grasp Hotel Norm - Normalization Queue. This pie chart shows the status of the hotel normalization queue. Hotels are in either in a “Normalized”, “Queued” or “NotFound” status. You can drill into the “NotFound” pie slice and research those hotels within your back office system.
- Line - Grasp Hotel Norm - Hotel Properties Normalized. This widget provides an overview of the normalization counts such as total hotels normalized, hotels normalized automatically via code (machine) and hotels normalized by a human.
Would Grasp Hotel Normalization “fix” the various names for a hotel property?
For example: Hilton Times Square, Hilton 42nd Street, New York Hilton Times Square? Yes. Grasp Hotel Normalization takes these various property aliases and collapses them down into a single master “Hilton Times Square 42nd Street New York” hotel property.
Which data points are normalized?
- Name
- Street
- City
- State
- Zip/Postal Code
- Country
- Phone Number
- Chain
- Latitude
- Longitude
Which data sources are normalized?
Back office, manual uploads and GDS data are supported. Grasp Hotel Normalization works across any data source that can provide at least 2 of the following four pieces of information: hotel name, hotel address and/or hotel phone number.
With what frequency is hotel information updated?
After each sync or data upload, new or changed hotels are detected and re-queued for normalization.
How long does it take to normalize my data?
When first onboarding with Grasp Hotel Normalization, an initial baseline list of your hotel properties are queued for normalization. This baseline can take between 7 and 10 days to normalize. After this baseline has been normalized, new or changed hotel properties are normalized within 1 day.
How do you maintain new names, new properties etc.?
The system detects new distinct hotel addresses, and if they have not previously been normalized, they will be queued for normalization. In the case of hotel properties being bought and sold between chains, Grasp Hotel Normalization is able to detect the change of ownership and normalize it accordingly.
How does Grasp handle corrections?
A QC report is generated and sent internally to a Grasp data entry team. The report captures any instances where the hotel normalization algorithm may not have normalized the hotel correctly. This report is reviewed and corrections will be applied during the next data import process (typically the next day).
How do the reports incorporate hotel normalization?
- When running a report that contains hotel data, there is a new toggle: “Use Normalized Hotel Data?” that a customer can toggle between Yes or No at report runtime or report schedule time.
- When set to Yes, the imported travel data is matched to our master hotel property table and normalized information about the hotel is return to the reports (hotel name, address, phone, chain code etc.).
- When set to No, the raw/un-normalized back office data is returned.
What data sources are normalized?
Back office, manual uploads and GDS data are supported for normalization. Grasp Hotel Normalization works across any data source that can provide at least 2 of the following four pieces of information: hotel name, hotel address and/or hotel phone number.
What about accuracy?
Grasp Hotel Normalization provides a quality control report that can be reviewed to determine current normalization accuracy. This report provides transparency with the normalization process and is reviewed by Grasp data entry staff as part of the product offering.
Is Grasp Hotel Normalization exclusively for properties in North America?
Grasp Hotel Normalization is for domestic and international properties.
Do I need to purchase other Grasp products in addition to Grasp Hotel Normalization?
Grasp Hotel Normalization is an add-on to GraspDATA, so you must have GraspDATA in order to utilize it.
Can GraspDATA be utilized during the time hotels are being normalized?
Yes. The Grasp Hotel Normalization process will not affect GraspDATA usage or performance.
What is the difference between Grasp Hotel Normalization data and the info in the back office?
Back offices are filled with hotel information either interfaced from the GDS or hand-entered. The same hotel can end up being entered multiple times with various spellings, addresses etc. This, in conjunction with combining your hotel data with other TMCs/data feeds (in the case of a consolidation) further compounds the issue. Grasp Hotel Normalization takes care of this duplication, and collapses the multiple aliases down into a master hotel property.
Do corrections and normalizations apply to only customer (DK) or all customers?
When hotels are normalized, either thru the code algorithms or manually, ALL Grasp customers and all corporate customers benefit from the normalized data.
How is Grasp Hotel Normalization installed?
It is an add-on to GraspDATA, and no additional software installation is required.
Implementation timeline and hours for training?
For existing GraspDATA customers, typically 2 weeks; for new customers, typically 4 weeks. Only 1 hour of training is needed, as the normalization is automated behind the scenes.
Will Grasp Hotel Normalization make my reports run slower?
No. The Grasp Hotel Normalization process runs offline and the calculations are updated offline. Normalization does not run during the sync/data import nor does it run during report runtime.
How do I interpret the QC report?
- The hotel normalization QC report is provided to show at a detailed level which hotel properties may not have been normalized correctly by the Grasp Hotel Normalization code algorithms. This report provides transparency and deeper insight into the state of the normalization process for your hotel data. Grasp’s data entry team members periodically review this report and make corrections or mark entries as “passing”. Each hotel property correction or entry marked as “passing” will fix associated bookings.
- Explanation of each tab within the QC report:
- Grasp Hotel Normalization_STATUS = overall summary of the state of the hotel normalization for the date period the report was ran. The key columns are E and H. offering the percentage of hotel properties that might be incorrectly normalized, and the # of bookings associated with those hotel properties.
- QC_ByDistance = this is a QC check where the raw geocoded address in comparison with the normalized address differs by a mileage threshold. Typically the geocoded raw address should be within the vicinity of the final hotel property chosen by the code algorithm and this tab provides those instances where the miles differs significantly.
- QC_ByHotelName = this is a QC check where the raw hotel name differs significantly from the normalized hotel name that the code algorithm selected (or the hotel chain differs significantly).
- QC_ZeroAccuracyNotFound = this is where the code algorithm could not arrive at a reasonable result and requires manual correction by a Grasp data entry team member.
Will Grasp Hotel Normalization update my back office too?
At this time, Grasp Hotel Normalization exclusively works with GraspDATA reporting.