Categories

Scoring my performance on the third fiscal quarter forecast

Did I get anything right in last forecast?

The table below shows the scores for my various estimates[1]

Last quarter I managed a B+ overall but this quarter was a C+. The iPad estimate was a complete failure mitigated slightly by a, directionally at least, correct call on the iPhone. The failure on iPad dragged down revenue and EPS while the accuracy on iPhone improved the gross margin. iPod and Mac estimates were consistent with recent performance.

The error on iPad is a harsh lesson which shows how the category is still unpredictable. With only four quarters where we could measure y/y growth, and with those quarters showing 183%, 166%, 111% and 151% growth, I assumed that the 150% growth was sustainable.

That turned out not to be the case. The iPad slowed to 84% unit growth and 51% revenue growth. It’s perhaps unrealistic to expect 150% growth for more than one year. This is the main learning for the quarter and I will adjust my expectations accordingly. Overall, the product has sold 84 million units since launching in early 2010. I now expect it will sell approximately 80 million this year and 150 million units in 2013.

Note: Grading algorithm is based on this table:

  • Danny

    Tracking your progress and deviation from actual results is good. Being close to actuals is desired, and far off is undesirable. But I don’t conceptually see how you can assign letter grades. Grades are reserved for circumstances (like school) where information is known and somebody is trying to learn it. The grade represents how well they have learned it, with an A being a match to knowledge recited, and an F being a failure to have ingested the knowledge. But you are looking at data that is unknown, fluctuating and unpredictable. It is so even for Apple. Scientists don’t give themselves a “Fail” when their experiment or search in the universe produced different results than expected. If anything, it may generate in them an epiphany, as more data and is accumulated, focusing their knowledge even further. Keep up the good work.

    • http://www.facebook.com/markgoldmd Mark Gold

      I have to agree. It is a frequent mantra that you can not use past performance of a Stock, mutual fund or … As a predictor of future performance. And with out continuous inside Apple information on sales, predicting sales is like reading tea leaves. Visiting Apple stores, hearing rumors from suppliers, looking at past performance and making educated guesses is not science. But unfortunately this is all we have. Hoarce do not take it personally, good work.

      • Chandra

        Mark: if you have been following Horace for a while, you would not have written that, since there are lot more information Horace derives from publicly available data.

    • Sander van der Wal

      If Horace has a *testable* hypothesis which he uses to predict the sales, he can grade himself on the outcome of that prediction.

      And it appears to me he has such a hypothesis. By looking at the amount of money spend on the logistical chain he can predict how much devices Apple is creating. And if Apple sells all devices they make he has also a reliable predictor of sales (or shipments).

    • actualbanker

      I think it’s useful. Since Horace is in a group (or class) of analysts it makes sense to have a simple proxy like a grade. The real usefulness would be when you have grades for all the other analysts and can compare. Over time the “better” analysts should have a better GPA. I bet Horace, despite a couple misses, is near the top. Until we can compare all of them over time it serves as a way for Horace to grade his own performance. “Am I getting better?” “Is my model more robust?” etc..

      • http://www.asymco.com Horace Dediu

        This is how the grading was initially envisioned: as a “report card” for the performance of all analysts. I begin with my own performance and then update a database for others. http://www.asymco.com/hire-me/apple-analyst-data/
        The idea of grades was so as to make the performance more easily understood. Raw measurements of error are more accurate.

    • Smitty

      Your basic idea is flawed in that education is not learning a specific body of content. That’s true if you’re learning a language, or medicine, or a specific craft, but many more subject areas in education teach critical thinking skills, problem solving, and creative approaches to problems with multiple solutions. Therefore, there is no “right” answer to many of the questions, without looking back from the point where the final product can be judged on the basis of history. If I am learning computer programming, say, getting an “A” would not only mean that I’ve mastered the language or systems under study (which would be a “C” – anyone coming out of the course should be able to do that), but that I am able to create more efficient, less resource-using programs, some of which may be better than what is contained in the body of current knowledge.
      Your thinking is usually the reason many of the most creative people don’t do well in traditional schools: they see better answers for the questions than the person(s) who made the tests, so don’t input the “correct” answer.

    • Smitty

      Your basic idea is flawed in that education is not learning a specific body of content. That’s true if you’re learning a language, or medicine, or a specific craft, but many more subject areas in education teach critical thinking skills, problem solving, and creative approaches to problems with multiple solutions. Therefore, there is no “right” answer to many of the questions, without looking back from the point where the final product can be judged on the basis of history. If I am learning computer programming, say, getting an “A” would not only mean that I’ve mastered the language or systems under study (which would be a “C” – anyone coming out of the course should be able to do that), but that I am able to create more efficient, less resource-using programs, some of which may be better than what is contained in the body of current knowledge.
      Your thinking is usually the reason many of the most creative people don’t do well in traditional schools: they see better answers for the questions than the person(s) who made the tests, so don’t input the “correct” answer.

    • Smitty

      Your basic idea is flawed in that education is not learning a specific body of content. That’s true if you’re learning a language, or medicine, or a specific craft, but many more subject areas in education teach critical thinking skills, problem solving, and creative approaches to problems with multiple solutions. Therefore, there is no “right” answer to many of the questions, without looking back from the point where the final product can be judged on the basis of history. If I am learning computer programming, say, getting an “A” would not only mean that I’ve mastered the language or systems under study (which would be a “C” – anyone coming out of the course should be able to do that), but that I am able to create more efficient, less resource-using programs, some of which may be better than what is contained in the body of current knowledge.
      Your thinking is usually the reason many of the most creative people don’t do well in traditional schools: they see better answers for the questions than the person(s) who made the tests, so don’t input the “correct” answer.

    • Smitty

      Your basic idea is flawed in that education is not learning a specific body of content. That’s true if you’re learning a language, or medicine, or a specific craft, but many more subject areas in education teach critical thinking skills, problem solving, and creative approaches to problems with multiple solutions. Therefore, there is no “right” answer to many of the questions, without looking back from the point where the final product can be judged on the basis of history. If I am learning computer programming, say, getting an “A” would not only mean that I’ve mastered the language or systems under study (which would be a “C” – anyone coming out of the course should be able to do that), but that I am able to create more efficient, less resource-using programs, some of which may be better than what is contained in the body of current knowledge.
      Your thinking is usually the reason many of the most creative people don’t do well in traditional schools: they see better answers for the questions than the person(s) who made the tests, so don’t input the “correct” answer.

    • Smitty

      Your basic idea is flawed in that education is not learning a specific body of content. That’s true if you’re learning a language, or medicine, or a specific craft, but many more subject areas in education teach critical thinking skills, problem solving, and creative approaches to problems with multiple solutions. Therefore, there is no “right” answer to many of the questions, without looking back from the point where the final product can be judged on the basis of history. If I am learning computer programming, say, getting an “A” would not only mean that I’ve mastered the language or systems under study (which would be a “C” – anyone coming out of the course should be able to do that), but that I am able to create more efficient, less resource-using programs, some of which may be better than what is contained in the body of current knowledge.
      Your thinking is usually the reason many of the most creative people don’t do well in traditional schools: they see better answers for the questions than the person(s) who made the tests, so don’t input the “correct” answer.

  • tedcranmore

    I think your method is too lenient on the Gross Margin grading. This is such a key number and missing by 200 points isn’t really in the “A” class if you ask me. My thought would be to measure “GM differential from Apple guidance”. I’m actually surprised given your good estimate of iPhones and your surprising number for iPad, you really should have had a significantly GM with that mix.

  • judsontwit

    Are you aggregating your previous forecast advice from other quarters? It’d be interesting to see a composite of how your strategy has changed. :)

  • http://www.facebook.com/ima.dumbash Ima Dumbash

    As my name indicates, I’m not too smart. I am also fairly new to your blog. What I would like to know is how you determine your estimates. Do you do channel checks? Do you have contacts at suppliers? What basis do you use to estimate each products sales number? If you have no data with which to base your estimates then how can you grade yourself?
    My apologies if you have addressed this before. But I am trying to get an understanding into your methodologies. Thx

    • http://www.asymco.com Horace Dediu

      I don’t do channel checks. I don’t have contact with any supplier. I use analysis and inference and available public information. There is a great of deal of public data to base estimates.

    • judsontwit

      Don’t forget to check the two years of back posts. That’s a lot of methodology publicly chronicled.

    • judsontwit

      Don’t forget to check the two years of back posts. That’s a lot of methodology publicly chronicled.

  • Chandra

    What will be the grade for the ultimate insider of all, Apple? Assuming their guidance is their prediction?

    • http://www.asymco.com Horace Dediu

      Quite poor, but it’s not used for two reasons: 1. the grading would have to be on only earnings and revenues and gross margin. Apple does not guide on shipments. 2. Guidance is not designed to offer an accurate prediction. It’s designed to protect management from lawsuits (among other things.)

      On the other hand, most professional analysts don’t offer estimates with the intention of being accurate. They are in the business of selling a rating on a stock and their estimates are designed to support a rating (buy, sell, etc.)

    • http://www.asymco.com Horace Dediu

      Quite poor, but it’s not used for two reasons: 1. the grading would have to be on only earnings and revenues and gross margin. Apple does not guide on shipments. 2. Guidance is not designed to offer an accurate prediction. It’s designed to protect management from lawsuits (among other things.)

  • mshipe

    The sales rate for the iPad is unprecedented as an Apple product but it also had the benefit of an unprecedented number of distributors/retailers (e.g. Wal-Mart, Target, Best Buy). Since iPhone sales growth seems to be correlated with the introduction of new distributors, what will be the best manner to project iPad sales growth?