Model 3 Delivery Estimator

Troy

Well-Known Member
Joined
Sep 18, 2017
Messages
385
Location
London, UK
Country
Country
#1
I created this tool in September 2016. I wanted to start this thread to keep the discussion separated from the invites spreadsheet. Here is the link: Model 3 Delivery Estimator

Here are the weekly production rates it uses. This chart is linked to the Estimator and it will auto-update to show the current settings.
 
Last edited:

Troy

Well-Known Member
Joined
Sep 18, 2017
Messages
385
Location
London, UK
Country
Country
#2
What's with the Model 3 Delivery Estimator spreadsheet? Has it become hyper-dynamic?
The changes are related to Tesla prioritizing Canada deliveries. There is conflicting information on how serious Tesla is about trying to extend federal tax credits until the end of 2018. Tesla.com's Canada estimates suggest a loose plan that won't actually work. For example, if you look at PaulN's estimates on the invites spreadsheet, he reports Late 2018, Early 2019, Late 2018 while everybody else from Canada who reserved before Paul report Mid 2018, Early 2019, Mid 2018. If Tesla was serious about the Canada idea, why would they not fulfill all Canada orders they can to reduce the excess? They will have too much excess at the end of Q2.

However, there are also some clues that suggest Tesla might be serious about this idea. If you go to tesla.com and look at the Model X configurator, it shows June 2018 if you order now. That delay looks intentional. In addition, in the Ride the Lightning podcast on 18 Feb, Ryan said he heard from a trustworthy source that Tesla's Canada play is about the federal tax credits.

I moved from May 2019 to October 2018 to now January 2019 again.
There was a major update on 17th Feb 2018. The estimates between 8-17 February were not very good.
 

Bokonon

Self-identified Teslaholic
Moderator
TOO Supporting Member
Joined
Apr 12, 2017
Messages
3,040
Location
Boston
Tesla Owner
Model 3
Country
Country
#3
I created this tool in September 2016. There is a thread in the other forum here but I wanted to start a thread here as well to keep the discussion separated from the invites spreadsheet.
Great work as always, and thanks for starting this thread!

I'm a big fan of this "version 2.0" of the estimator for a variety of reasons. The production modeling in the "Data" tab seems a lot more flexible (which presumably would make it easier to tweak the estimates in response to new data points). Also, the ability to compare the calculations for two different configurations side-by-side is an especially nice touch!

I have a couple of questions about what's on the Data tab:
  • It looks like you have P and D starting production at the same time and ramping up together, first in Canada and then in the U.S. This is a change from earlier versions of the spreadsheet, which had D and P starting production a few months apart. Out of curiosity, is this change based on some (or all) of the data points below, or is there something else that is leading you to believe that P and D will go to production together?
    • Elon's tweet from last July that P production would start "middle of next year"
    • The statement from the last quarterly earnings call that Tesla will re-prioritize production around higher-margin configurations to improve overall margins for the Model 3.
    • Speculation that the Model 3 with red brake calipers spotted doing launches on the Fremont test track, as well as Franz's personal Model 3, are both PD variants
  • If I'm reading the Data tab correctly, it looks you have D production starting out with 466 produced in the first week, 503 in the second week, and ramping up from there. (Similarly, P production starts out with 93 cars in the first week.) Given that Tesla has previously stated that the production curve for the D and P variants will resemble the initial S curve for the First Production configuration, shouldn't the P and D production numbers start out lower and increase at a slower rate? In other words, shouldn't columns N and O in April look more like Column H in, say, September or October?
Thanks again for everything you've done to maintain this tool and keep it relevant, useful and even entertaining in the face of Tesla's constantly-shifting positions and timelines. :)
 

Troy

Well-Known Member
Joined
Sep 18, 2017
Messages
385
Location
London, UK
Country
Country
#4
Hi, @Bokonon.

1. Model 3 LRD and P start times: This is related to the second option you mentioned about prioritizing high-profit margin cars. Delaying standard range must have been a difficult decision because Elon said they wanted to avoid complexity because if they had introduced AWD, more things could go wrong when they are trying to ramp up. Now they have ditched this plan.

2. D and P variants resembling the S curve: In the current model, both of these start out slowly. If you look at cells H18:J18 on the data tab, it shows what percentage of production is allocated to that region & trim combination. I'm considering analyzing Tesla.com estimates in more detail. Tesla.com AWD estimates show mid-2018 for early reservations and late-2018 for late reservations. It is actually possible to calculate percentages from the timing it changes from mid to late. The tricky part is to automate that calculation instead of doing it manually.
 

OrangeJulius

Active Member
Joined
Jan 19, 2018
Messages
188
Location
San Francisco, California
Tesla Owner
Model 3
Country
Country
#5
Sorry, I'm still not getting why the estimator is, for me and perhaps others, adding an additional 10 days delay to delivery every day.

This Sunday, the 18th, it estimated delivery on 5/17/18, then on Monday it estimated 5/27/18, now today it's estimating 6/6/18.

The only data points that have changed have been the "Global prod sequence no". "Configuration" date and "Production" date.

Why are these data points changing from one day to the next (and in a negative direction)?

p.s. If I were waiting for the SR, it's estimating delivery of 4/17/19 - over three years after placing the original reservation. :mad:
 

Troy

Well-Known Member
Joined
Sep 18, 2017
Messages
385
Location
London, UK
Country
Country
#7
@OrangeJulius, your estimates have changed because I moved Canada production sooner and increased the percentage which means less production for the US around April, May, and June 2018.

The first half of the screenshot shows the current settings. 5%+20%+4% of Tesla's production between Apr 1st and May 1st is allocated to Canada. In the previous version, this period was zero and the US was 100%.

I'm adjusting these percentages based on how serious Tesla's Canada play appears to be. If they are serious, all Canada production must happen by the end of June 2018. Actually, to be precise, by June 15th. That might be the next adjustment.

In case you are not following developments, here is a summary: When Tesla reaches 200,000 US sales, it will trigger the phaseout calendar for federal tax credits. If they hit 200K in Q2, $7,500 will last until 30th Sep 2018. If Q3 then after 31st Dec 2018. At the end of 2017, they were around 162K. Avoiding 200K in Q2 is extremely difficult. Therefore I'm not sure about this idea but I'm watching developments.


 
Last edited:

Bokonon

Self-identified Teslaholic
Moderator
TOO Supporting Member
Joined
Apr 12, 2017
Messages
3,040
Location
Boston
Tesla Owner
Model 3
Country
Country
#8
Sorry, I'm still not getting why the estimator is, for me and perhaps others, adding an additional 10 days delay to delivery every day.
This Sunday, the 18th, it estimated delivery on 5/17/18, then on Monday it estimated 5/27/18, now today it's estimating 6/6/18.
The only data points that have changed have been the "Global prod sequence no". "Configuration" date and "Production" date.
Why are these data points changing from one day to the next (and in a negative direction)?
Short answer: because the modeling in the estimator spreadsheet itself has changed a lot over the past few days.

Prior to yesterday, when he started this thread, Troy typically announced most changes and updates to the spreadsheet in the Delivery Estimator thread on TMC that he linked to above. Late last week, he released a major update to the spreadsheet which completely changed how his model performed its calculations, and most people's estimates changed significantly as a result. Since then, he has made additional tweaks to the model (most recently, yesterday at about 3pm PST), which again caused most people's estimates to change.

I have no doubt Troy will continue to make adjustments as new data points come to light and the model's assumptions are either confirmed or disproved. After all, that's exactly how any data-driven forecast should work. :)

And keep in mind, that's all this spreadsheet is: a forecasting tool. Much like a weather forecast, it distills a complex and chaotic process with a large number of unknown variables into a simplified, understandable representation of what will happen if certain assumptions and logic hold true. It has no inherent authority, nor does it have any influence over the events it attempts to predict, nor does it provide any guarantee of how accurate its estimates will ultimately prove to be. You could take delivery on May 17th, May 27th, or June 6th... or you could take delivery on April 30th or July 2nd. At this point, there are simply too many unknowns to make a prediction that precise.

I would focus instead on the *range* of the last few predictions (i.e. mid-May to early-June), compare them to Tesla's "official" delivery window, and center your expectations in the middle of the two.

EDIT: Looks like Troy just responded with the specific changes he made in this latest iteration. Hopefully we'll get some clarity on the divert-production-to-Canada scenario soon, because your estimated delivery unfortunately falls right in the middle of when that scenario would hypothetically play out... i.e., you get bonus uncertainty! :)
 

Troy

Well-Known Member
Joined
Sep 18, 2017
Messages
385
Location
London, UK
Country
Country
#9
You can add the Model S to that statement now.
Thanks for the info. I think I know why it says June instead of July. I think the configurator actually displays production times. US production can resume at full speed after 15 June 2018 because cars that enter production after this date won't be delivered in Q2. The 200K count only includes cars that are delivered.

I'm considering reducing Q2 production similar to what is happening in Q1. If you look at the chart in the opening message, it shows a slow increase until the end of February and then steep increase in March. If we do the same for Q2, I think I can make the numbers work for federal tax credits.
 
Last edited:

Carl_P

Active Member
Joined
Jan 30, 2018
Messages
30
Location
Philadelphia, PA
Tesla Owner
Model 3
Country
Country
#10
Sorry, I'm still not getting why the estimator is, for me and perhaps others, adding an additional 10 days delay to delivery every day.

This Sunday, the 18th, it estimated delivery on 5/17/18, then on Monday it estimated 5/27/18, now today it's estimating 6/6/18.

The only data points that have changed have been the "Global prod sequence no". "Configuration" date and "Production" date.

Why are these data points changing from one day to the next (and in a negative direction)?

p.s. If I were waiting for the SR, it's estimating delivery of 4/17/19 - over three years after placing the original reservation. :mad:
I ran my Estimator last week and got delivery 9/2018. I ran it today and got 4/1/19.

It seems the SR people just got majorly screwed. All speculation of course. Tesla.com is still telling me "Late 2018"
 

Troy

Well-Known Member
Joined
Sep 18, 2017
Messages
385
Location
London, UK
Country
Country
#11
Calculations have been updated. To do that, I have processed Tesla.com estimates as reported on the invites spreadsheet. Here are some charts I created today. You can find these on the Delivery tab here. These charts turned out to be very useful because now I don't have to guess how to balance the percentages.

For example, it shows 66% of US non-owners are expected to get their car in Late 2018 and the rest in Early 2019. I tried to match these as close as possible. In this case, I have 60% instead of 66% because my production estimates are slightly lower than Tesla's. I have increased the cancellation rate slightly and reduced the owner percentage slightly. I have done more research on all of these issues. I think the Estimator is now more accurate than it has ever been.



 
Joined
Aug 16, 2017
Messages
13
Location
Irving
Country
Country
#13
I hope that's true; our delivery is now almost 3 weeks sooner than the last Estimator calculation. :)
I tried out the latest Estimator and also found that the delivery is now estimated to be 3 weeks earlier. @Troy, Is the biggest factor driving the advanced delivery dates due to the fact that the owner percentages have been lowered and cancellation rate increased (based on recent actual data)?
 

Troy

Well-Known Member
Joined
Sep 18, 2017
Messages
385
Location
London, UK
Country
Country
#14
@tommythewho, no, the biggest factor for you was Canada deliveries. I think your estimate moved from July to June. This is because I have reduced deliveries to Canada in April, May, and June because Tesla.com shows Late-2018 for 10% of Canada reservations. More info about Canada adjustment here.
 

Frank99

Model 3 owner since May 3, 2018
TOO Supporting Member
Joined
Aug 3, 2017
Messages
250
Location
Phoenix, AZ
Tesla Owner
Model 3
Country
Country
#15
Troy - great work; especially since it estimates that I'll get invited to configure this week! (But I'll bet you a quarter that I don't).

One question - my "Global Reservation Number" is listed as #14215, but back in 2016 when the common_reservation_id was available, I figured it as being somewhere around #36000. I'm guessing that the discrepancy is that I didn't take into account time zones - I considered someone in New York who reserved two hours after store opening was an hour ahead of me (reserved an hour after store opening in Arizona), where I'm guessing you're following what Tesla seems to be doing now - prioritizing based on time after store opening, not absolute time. In this case, my priority would be similar to someone who reserved an hour after local store opening in New York, or Ohio, or California. Is this a good guess?

Do you have the common_reservation_id spreadsheet from way back then? If not, I have a cleaned up copy that I can send you.
 

PTC Gator

Active Member
Joined
Sep 2, 2017
Messages
163
Location
Atlanta, GA
Tesla Owner
Model 3
Country
Country
#16
@tommythewho, no, the biggest factor for you was Canada deliveries. I think your estimate moved from July to June. This is because I have reduced deliveries to Canada in April, May, and June because Tesla.com shows Late-2018 for 10% of Canada reservations. More info about Canada adjustment here.
Is there a way to just have a permanent "like/informative" selection for Troy? Pretty much everything I've seen from him or KarenREI is useful and informative. Certainly there are others, but those two come to mind for me.
 

Troy

Well-Known Member
Joined
Sep 18, 2017
Messages
385
Location
London, UK
Country
Country
#17
I'm guessing that the discrepancy is that I didn't take into account time zones
Yes, that's the reason. Imagine the following are in store reservations on day 1:

10 am GMT = 2 am PST John
10 am ET= 7 am PST Elizabeth
10 am PST = 10 am PST Robert

In this scenario, Robert's global reservation number would be higher than Elizabeth's. However, it would be unfair if Elizabeth received her car sooner because her time zone happens to be sooner. Therefore I eliminated the time zone differences for in-store reservations on day 1. All 3 are equal. When you enter 31st, in-store, you can select any time zone you want and your global reservation number won't change, except when the time gets close to online reservations.
 
Last edited:

Troy

Well-Known Member
Joined
Sep 18, 2017
Messages
385
Location
London, UK
Country
Country
#19
Hi, @ChrisH. Currently, the estimator uses 14.6% cancellation rate (85.4% active reservations rate). Here are some data points:
  • 96.8% on 15 May 2016. This is based on the first quote below by Tesla. The 373,000 number released by Tesla was after 8,000 cancellations and 4,200 duplicates were removed. Before these were removed, the number would have been 373,000+8,000+4,200= 385,200. That means, 373,000/385,200= 96.8% active reservations on 15 May 2016.
  • 87.8% on 28 July 2017. This is based on the second quote below by Tesla. 455/518= 87.8% active reservations.
  • 85.4% is my current estimate.
Tesla: as of May 15, 2016, we held deposits from about 373,000 customers who had made reservations for this car. This reservation total is a net number after customer cancelations of about 8,000 and after about 4,200 reservations that we canceled ... Source
Tesla: ...there have been 518,000 gross reservations for Model 3 and then we have 455,000 net reservations. But those cancellations occurred over the course of more than a year. Source
I reduced it from 87.8% to 85.4% because I thought some people who want the standard range version might cancel after it was pushed to early 2019. However, this wouldn't have a big overall effect because most buyers want the long-range version.
 
Last edited: