New UA Upgrade Success/Failure Tracking Site
#16
Join Date: May 2009
Location: Washington, DC
Programs: UA 1K 1MM, AA, DL
Posts: 7,418
It does need to be updated to reflect at least new status levels (no plat. on it) and also new routes.
My preference is to go that way, but if the owner can't be reached . . .
(ETA: I sent an email to the person listed as the owner on Whois)
Last edited by drewguy; Jan 23, 2013 at 11:44 am
#17
FlyerTalk Evangelist
Join Date: Sep 2008
Location: Jersey Shore/YYZ
Programs: UA 1K, Marriott Plat, Hilton Diamond, Hertz PC
Posts: 12,521
While I agree that compliance would be a problem, there are a couple other suggestions I have if you're going to go forward:
- When did the upgrade clear (T-120-96, T-96 to 48, T-48 to 24, T-24, at Gate)
- If not cleared, where were you on the upgrade list (not essential)
- If offered buy-up, what was the dollar amount.
Make as much optional as possible to still be useful.
It's too bad there's no tripit-like way to have people forward - like if they actually receive the premier upgrade notification.
Maybe you could write a greasemonkey script that would allow one-click submission to the site that captures basic flight information from united.com - click here if upgraded for example and then it prompts for the remaining info as suggested.
- When did the upgrade clear (T-120-96, T-96 to 48, T-48 to 24, T-24, at Gate)
- If not cleared, where were you on the upgrade list (not essential)
- If offered buy-up, what was the dollar amount.
Make as much optional as possible to still be useful.
It's too bad there's no tripit-like way to have people forward - like if they actually receive the premier upgrade notification.
Maybe you could write a greasemonkey script that would allow one-click submission to the site that captures basic flight information from united.com - click here if upgraded for example and then it prompts for the remaining info as suggested.
#18
Join Date: Oct 2006
Location: DCA, ex-IAH
Programs: nada
Posts: 1,368
I like it too. Also if there was room to add on a few other data points that I find useful:
Cabin size
# upgraded
# not upgraded
Status level: Nonrev (my brother cleared SLC-ORD at christmas a couple of years ago, with empty seats in F)
# empty seats in F (actually quite a few of my holiday time flights go out this way)
Notes? I wasn't going to clear on flight A so I moved to earlier flight B at T-18, and cleared 4 hours after that, IIRC. I'll likely be trying to repeat this feat tomorrow for Friday if it looks the same situation.
Cabin size
# upgraded
# not upgraded
Status level: Nonrev (my brother cleared SLC-ORD at christmas a couple of years ago, with empty seats in F)
# empty seats in F (actually quite a few of my holiday time flights go out this way)
Notes? I wasn't going to clear on flight A so I moved to earlier flight B at T-18, and cleared 4 hours after that, IIRC. I'll likely be trying to repeat this feat tomorrow for Friday if it looks the same situation.
#20
Original Poster
Join Date: Aug 2011
Programs: UA 1K
Posts: 8,634
Thanks for the feedback, everyone! Here are some responses to the issues that folks raised:
On the issue of compliance, I’m inclined to agree with drewguy – I think that some data is better than no data, and I think that flyertalk and prior use of udustats (from what I can tell, as I was never a user) shows that people are quite ready to log data about their upgrades (among other things). My understanding was that use of udustats had fallen off because things were all messed up during the merger, and the site hasn't been updated to reflect the new program.
I think that concern about data bias depends a bit on how people are using the data. If a Silver takes a look at unfiltered data made up almost entirely of 1K’s and Platinums in trying to determine her likelihood of getting an upgrade, she’s gonna be mistaken. But if people understand the nature of the pool, and the data is available in filtered form, it shouldn’t be so bad. I understand that some people here might have ways of ending up with an abnormally high upgrade percentage for their level of status (e.g., calling and complaining when they don’t appear on the upgrade list), I doubt that that has earth-shattering implications to the aggregate upgrade percentages. Maybe I’m underestimating flyertalk folks.
It’s true that you might not be able to find a record of people with your exact level of status, boarding flights between your exact origin and destination, at the exact time and on the exact day of the week that you’re flying. Such is the nature of sampled data. I do think that you can glean a lot from aggregated data – say, of people with your status, also flying a non-hub to hub short-haul route, on a heavy weekend day (fri/sun). I hope to build in options that make this easy on both the reporting and data display sides.
I would love to have someone build an awesome script-based system that automatically scrapes from emails or that can act as a browser plugin and get data automatically. Nonetheless, I chose to go with the simple Google form route for three reason: (1) I haven’t produced a significant amount of code since I was in eighth grade, (2) keeping this super-simple means I’ll actually get it done and data will start getting recorded, and (3) using simple platforms means a lot more people will have the ability to maintain it in the future, so it doesn’t go the way of udustats.
One way to encourage compliance would be to have username collection and then provide a way to keep track of flyertalk users’ success rates. Some people seem to like to report that info in their signatures.
I will absolutely try to design the form to make it as easy as possible to fill out the minimum info necessary to be useful, which should increase reporting. Everyone hates long, convoluted forms. I’ll be asking you all about ways to do that in a follow-up post.
I, too, think it would be great to see the old version revived, or to incorporate this reporting into another tool. I'm just working with what I've got. I also worry that an alternative might suffer from the same issue of not being updated.
Google forms has its drawbacks, including ugly spacing and a lack of features like pop-up calendars. I'll look a bit into how to remedy some of those.
As I mention above, I personally think that the flexibility and approachability of Google forms, including the ability to feed info directly into a publicly viewable spreadsheet with graphs and pivot tables and the capacity for maintenance by non-tech-people, makes up for its shortcomings.
One thing that's nice is that, regardless of how people type the date, Google seems to figure it out and put it in the spreadsheet in a nice formatted way.
Suggestions all noted -- I will definitely be making many improvements and additions to the form itself; this was just a very rough cut.
The problem with the last website was compliance. You simply needed people to contribute to the site. Once people stopped, it didn't work. Moreover, if only a certain segment contributed, it was not useful. You created a selection bias.
The key to this site is that *everybody* need to be included. You need a way to capture all the data, not just data from people who have the time, energy, and willingness to enter it, in order to make it accurate & useful.
The key to this site is that *everybody* need to be included. You need a way to capture all the data, not just data from people who have the time, energy, and willingness to enter it, in order to make it accurate & useful.
Make as much optional as possible to still be useful.
It's too bad there's no tripit-like way to have people forward - like if they actually receive the premier upgrade notification.
Maybe you could write a greasemonkey script that would allow one-click submission to the site that captures basic flight information from united.com - click here if upgraded for example and then it prompts for the remaining info as suggested.
It's too bad there's no tripit-like way to have people forward - like if they actually receive the premier upgrade notification.
Maybe you could write a greasemonkey script that would allow one-click submission to the site that captures basic flight information from united.com - click here if upgraded for example and then it prompts for the remaining info as suggested.
While the site is a great idea, when i went to check on a flight from bradley to someplace (don't remember where) months and months into it, there were no entries at all.
If you are basing your data off what the passengers are entering, you will never get enough people to take the time to put in the data to get meaningful results.
If you are basing your data off what the passengers are entering, you will never get enough people to take the time to put in the data to get meaningful results.
The simpler the process, the more compliant people will be about sharing data - making them access/open/update a spreadsheet won't work well, but if you created a very simple front-end form hosted from the site that allowed a user to add data to preset fields and submit, and that form updated the spreadsheet, you might have a better shot at collecting more data.
I think that concern about data bias depends a bit on how people are using the data. If a Silver takes a look at unfiltered data made up almost entirely of 1K’s and Platinums in trying to determine her likelihood of getting an upgrade, she’s gonna be mistaken. But if people understand the nature of the pool, and the data is available in filtered form, it shouldn’t be so bad. I understand that some people here might have ways of ending up with an abnormally high upgrade percentage for their level of status (e.g., calling and complaining when they don’t appear on the upgrade list), I doubt that that has earth-shattering implications to the aggregate upgrade percentages. Maybe I’m underestimating flyertalk folks.
It’s true that you might not be able to find a record of people with your exact level of status, boarding flights between your exact origin and destination, at the exact time and on the exact day of the week that you’re flying. Such is the nature of sampled data. I do think that you can glean a lot from aggregated data – say, of people with your status, also flying a non-hub to hub short-haul route, on a heavy weekend day (fri/sun). I hope to build in options that make this easy on both the reporting and data display sides.
I would love to have someone build an awesome script-based system that automatically scrapes from emails or that can act as a browser plugin and get data automatically. Nonetheless, I chose to go with the simple Google form route for three reason: (1) I haven’t produced a significant amount of code since I was in eighth grade, (2) keeping this super-simple means I’ll actually get it done and data will start getting recorded, and (3) using simple platforms means a lot more people will have the ability to maintain it in the future, so it doesn’t go the way of udustats.
One way to encourage compliance would be to have username collection and then provide a way to keep track of flyertalk users’ success rates. Some people seem to like to report that info in their signatures.
I will absolutely try to design the form to make it as easy as possible to fill out the minimum info necessary to be useful, which should increase reporting. Everyone hates long, convoluted forms. I’ll be asking you all about ways to do that in a follow-up post.
That would be ideal, but its creator has gone missing from flyertalk.
It does need to be updated to reflect at least new status levels (no plat. On it) and also new routes.
My preference is to go that way, but if the owner can't be reached . . .
(eta: I sent an email to the person listed as the owner on whois)
It does need to be updated to reflect at least new status levels (no plat. On it) and also new routes.
My preference is to go that way, but if the owner can't be reached . . .
(eta: I sent an email to the person listed as the owner on whois)
Just tried filling out the form for a test. Everything looks great, could maybe use less spacing to make scrolling less necessary.
Also, i would encourage a standard date and time format, rather than have the user manually input that info. (the way it's set up now, some people will put in january 20th, 2013, while others will put in 1/20/13, while still others will do 20-jan, 13.) if you could use a calendar pop-up, that would be ideal.
Also, i would encourage a standard date and time format, rather than have the user manually input that info. (the way it's set up now, some people will put in january 20th, 2013, while others will put in 1/20/13, while still others will do 20-jan, 13.) if you could use a calendar pop-up, that would be ideal.
As I mention above, I personally think that the flexibility and approachability of Google forms, including the ability to feed info directly into a publicly viewable spreadsheet with graphs and pivot tables and the capacity for maintenance by non-tech-people, makes up for its shortcomings.
One thing that's nice is that, regardless of how people type the date, Google seems to figure it out and put it in the spreadsheet in a nice formatted way.
While i agree that compliance would be a problem, there are a couple other suggestions i have if you're going to go forward:
- when did the upgrade clear (t-120-96, t-96 to 48, t-48 to 24, t-24, at gate)
- if not cleared, where were you on the upgrade list (not essential)
- if offered buy-up, what was the dollar amount.
- when did the upgrade clear (t-120-96, t-96 to 48, t-48 to 24, t-24, at gate)
- if not cleared, where were you on the upgrade list (not essential)
- if offered buy-up, what was the dollar amount.
I like it too. Also if there was room to add on a few other data points that i find useful:
Cabin size
# upgraded
# not upgraded
status level: Nonrev (my brother cleared slc-ord at christmas a couple of years ago, with empty seats in f)
# empty seats in f (actually quite a few of my holiday time flights go out this way)
notes? I wasn't going to clear on flight a so i moved to earlier flight b at t-18, and cleared 4 hours after that, iirc. I'll likely be trying to repeat this feat tomorrow for friday if it looks the same situation.
Cabin size
# upgraded
# not upgraded
status level: Nonrev (my brother cleared slc-ord at christmas a couple of years ago, with empty seats in f)
# empty seats in f (actually quite a few of my holiday time flights go out this way)
notes? I wasn't going to clear on flight a so i moved to earlier flight b at t-18, and cleared 4 hours after that, iirc. I'll likely be trying to repeat this feat tomorrow for friday if it looks the same situation.
#21
Join Date: Jan 2010
Location: Aussie in ORD
Programs: Marriott Plat, Ua Gold, GE.. Sucker for punishment
Posts: 4,237
Let me add my support! I often check udustats and would love to see a reboot..
I can code some python if you need.. Make a github repo and I can fork and contribute
I can code some python if you need.. Make a github repo and I can fork and contribute
#22
Join Date: May 2009
Location: Washington, DC
Programs: UA 1K 1MM, AA, DL
Posts: 7,418
Date can be coded to day of week, but that should show up somehow.
I would also consider not using "insider" plane types (772, etc.). Instead, used the actual model number, and then whatever extension is necessary to distinguish it. Even better if set so when choosing, eg, 737, a submenu for types will show up--that will allow people who don't know the variants to still input the info.
You might combine RPU and GPU--aren't the two "equal" for purposes of upgrades, other than limits on where RPUs can be used and fare classes for GPUs?
No need to have "UA" at the beginning of the flight number.
#23
Join Date: Dec 2012
Location: WAS
Programs: AA PLT, Honors Diamond, Global Entry
Posts: 477
This is a great idea; I strongly support it. And if you launch a page on a crowdfunding platform (e.g. IndieGoGo), I'll happily throw down some cash.
An even better idea, however, would be to scrape United.com upgrade data and make it available on a public web database. The number of successful upgrades - and pax on the upgrade waitlist - is publicly available for every UA flight several days out. So if you could figure out how to scrape this data and present it in a usable format, anybody booking a ticket could easily figure out the historical average number of upgrades requested/confirmed on any particular UA flight or routing
To be sure, with this method you wouldn't know how many instruments were used on a flight versus CPUs, or the distribution of status pax. But using rough estimates based on public data (e.g. 2005 UA "elite pyramid" stats) you should be able to generate some pretty robust information about which routes/flights/days tend to offer the greatest CPU opportunities.
An even better idea, however, would be to scrape United.com upgrade data and make it available on a public web database. The number of successful upgrades - and pax on the upgrade waitlist - is publicly available for every UA flight several days out. So if you could figure out how to scrape this data and present it in a usable format, anybody booking a ticket could easily figure out the historical average number of upgrades requested/confirmed on any particular UA flight or routing
To be sure, with this method you wouldn't know how many instruments were used on a flight versus CPUs, or the distribution of status pax. But using rough estimates based on public data (e.g. 2005 UA "elite pyramid" stats) you should be able to generate some pretty robust information about which routes/flights/days tend to offer the greatest CPU opportunities.
#24
Original Poster
Join Date: Aug 2011
Programs: UA 1K
Posts: 8,634
Okay, people seem interested enough, so let's do this.
Moving on to the next step:
Some questions about some of the suggestions I've gotten:
These are great points.
For plane types, is there some definitive way of referring to different versions that people recognize? Something better than SeatGuru's 777-200 V6?
Not sure on the RPU/GPU issue. Is there any effect on priority?
When and why do downgrades happen?
I assume that you mean numbers upgraded and not upgraded based on the lists at the gate?
Are nonrevs in a particular booking class?
I appreciate both of the offers of support and the ideas. Sadly, my job is a bit too full time to take on real coding or something I could justify asking people to support financially, but I'm hoping that maybe once this gets (back) off the ground, someone might go crazy with it.
Some edits made, many more to come.
Moving on to the next step:
- Conceptual stage.
- Question and form design.
- Data presentation design.
- Ongoing use and improvement.
Some questions about some of the suggestions I've gotten:
This is the most important one that's missing. Also add "at booking" and >120 hours.
Date can be coded to day of week, but that should show up somehow.
I would also consider not using "insider" plane types (772, etc.). Instead, used the actual model number, and then whatever extension is necessary to distinguish it. Even better if set so when choosing, eg, 737, a submenu for types will show up--that will allow people who don't know the variants to still input the info.
You might combine RPU and GPU--aren't the two "equal" for purposes of upgrades, other than limits on where RPUs can be used and fare classes for GPUs?
No need to have "UA" at the beginning of the flight number.
Date can be coded to day of week, but that should show up somehow.
I would also consider not using "insider" plane types (772, etc.). Instead, used the actual model number, and then whatever extension is necessary to distinguish it. Even better if set so when choosing, eg, 737, a submenu for types will show up--that will allow people who don't know the variants to still input the info.
You might combine RPU and GPU--aren't the two "equal" for purposes of upgrades, other than limits on where RPUs can be used and fare classes for GPUs?
No need to have "UA" at the beginning of the flight number.
For plane types, is there some definitive way of referring to different versions that people recognize? Something better than SeatGuru's 777-200 V6?
Not sure on the RPU/GPU issue. Is there any effect on priority?
I like it too. Also if there was room to add on a few other data points that i find useful:
Cabin size
# upgraded
# not upgraded
status level: Nonrev (my brother cleared slc-ord at christmas a couple of years ago, with empty seats in f)
# empty seats in f (actually quite a few of my holiday time flights go out this way)
notes? I wasn't going to clear on flight a so i moved to earlier flight b at t-18, and cleared 4 hours after that, iirc. I'll likely be trying to repeat this feat tomorrow for friday if it looks the same situation.
Cabin size
# upgraded
# not upgraded
status level: Nonrev (my brother cleared slc-ord at christmas a couple of years ago, with empty seats in f)
# empty seats in f (actually quite a few of my holiday time flights go out this way)
notes? I wasn't going to clear on flight a so i moved to earlier flight b at t-18, and cleared 4 hours after that, iirc. I'll likely be trying to repeat this feat tomorrow for friday if it looks the same situation.
Are nonrevs in a particular booking class?
This is a great idea; I strongly support it. And if you launch a page on a crowdfunding platform (e.g. IndieGoGo), I'll happily throw down some cash.
An even better idea, however, would be to scrape United.com upgrade data and make it available on a public web database. The number of successful upgrades - and pax on the upgrade waitlist - is publicly available for every UA flight several days out. So if you could figure out how to scrape this data and present it in a usable format, anybody booking a ticket could easily figure out the historical average number of upgrades requested/confirmed on any particular UA flight or routing
To be sure, with this method you wouldn't know how many instruments were used on a flight versus CPUs, or the distribution of status pax. But using rough estimates based on public data (e.g. 2005 UA "elite pyramid" stats) you should be able to generate some pretty robust information about which routes/flights/days tend to offer the greatest CPU opportunities.
An even better idea, however, would be to scrape United.com upgrade data and make it available on a public web database. The number of successful upgrades - and pax on the upgrade waitlist - is publicly available for every UA flight several days out. So if you could figure out how to scrape this data and present it in a usable format, anybody booking a ticket could easily figure out the historical average number of upgrades requested/confirmed on any particular UA flight or routing
To be sure, with this method you wouldn't know how many instruments were used on a flight versus CPUs, or the distribution of status pax. But using rough estimates based on public data (e.g. 2005 UA "elite pyramid" stats) you should be able to generate some pretty robust information about which routes/flights/days tend to offer the greatest CPU opportunities.
Some edits made, many more to come.
Last edited by iluv2fly; Jan 23, 2013 at 10:45 pm Reason: merge
#25
Join Date: May 2009
Location: Washington, DC
Programs: UA 1K 1MM, AA, DL
Posts: 7,418
Here's my suggestion of importance:
Must have:
- city-pair
- date/time
- status
- whether upgrade cleared
Nice to have:
- fare class (not everybody focuses on this, and it's second tiebreaker)
- flight number (these change a lot, so less useful)
- time before flight upgrade cleared (useful to know, but bottom line is whether you cleared)
- type of upgrade (possibly could put miles/rpu/gpu all in same bucket because they're supposed to be same priority)
- plane type (for some routes, it's often the same type of plane so not always relevant)
Optional
- waitlist rank/total at gate (gives some sense, but might be better in notes for those not upgraded. Hard to know the number upgraded because of people paying for F/C and upgrades that happen in advance--useless to see how many cleared at gate)
- downgrades (happen rarely and tracking it is a separate issue--perhaps leave to notes)
- non-rev passenger upgrades (limited audience. Could add as a option in fareclass)
- whether cash upgrade was offered (it's an interesting question, but not very useful to figuring out upgrade chances)
For plane types, is there some definitive way of referring to different versions that people recognize? Something better than SeatGuru's 777-200 V6?
737s might be more complicated (or maybe I just don't know all the variants).
#26
Join Date: Aug 2012
Programs: 2013: UA 1K (2013: 4/7 UDU, 2012: from GM to 1K!), Global Entry
Posts: 66
Contact the openflights.org author and add a feature request (better yet, it's open source, and I am happy to help too). I would much rather have the forms added to that system then use "yet another site".
#27
Join Date: Apr 2007
Location: SEA
Programs: UA Plat, AS MVP, HH Gold, MR Gold
Posts: 1,848
I am very curious about who's up in F, now that it's no longer me (without a meaningful change in my flying behavior). I would definitely contribute and use the data.
I would add to upgrade types YB(M)-elite upgrades, buy-up at booking and buy-up at check-in (rather than just asking whether there was an offer). My hunch is that is where all the CPUs have gone.
I would add to upgrade types YB(M)-elite upgrades, buy-up at booking and buy-up at check-in (rather than just asking whether there was an offer). My hunch is that is where all the CPUs have gone.
#28
Join Date: May 2009
Location: Washington, DC
Programs: UA 1K 1MM, AA, DL
Posts: 7,418
In my mind, the objective is to inform other flyers of what flights are more or less likely to result in upgrades.
Information about buy-ups, Y/B/M upgrades, etc. don't really inform that question. Rather, they are, as you say, providing information about who is in the F seats. But, even so, they are incomplete because completeness would require us also to ask about paid F tickets. What's more, they won't provide a "rate" of anything terribly useful, unless the goal is to figure out what the likelihood of being offered a TOD upgrade is.
#29
Join Date: Apr 2007
Location: SEA
Programs: UA Plat, AS MVP, HH Gold, MR Gold
Posts: 1,848
I agree that defining a purpose is useful, and that the objective is to help us plan for what flights are more likely to result in upgrades. But I'd argue that, in addition to helping us understand who is in the F seats, knowing about buy-ups and YBMs advances that cause. In particular, it helps answer the question of, if I am on a particular flight, or set of flights, what instrument do I need to use to have a shot at an upgrade? On some flights, CPUs will work for a 1K like me. On others, I might need to apply an instrument. On others, instruments might not work well, so then I might consider buying (and possibly wait listing) an M-up, or a Y-up. On a busier flight, I might need to realize that I need to take the buy-up offer, because that's my only shot. And on ORD-SFO on Monday morning, I might need to know that buying F is my only realistic option.
#30
A FlyerTalk Posting Legend
Join Date: Apr 2001
Location: PSM
Posts: 69,232
Self-reporting will never have a sufficient quantity of data to allow for good analytics to happen.