Discussion: View Thread

This relates to two ASA-related eGroup postings I've seen recently:

(1) the discussion within the Consulting Section eGroup about working under retainer arrangements.

(2) the announcement of the next (68th) Annual Deming Conference on Applied Statistics, December 2-7, 2012; Atlantic City, NJ. The full program is available at www.demingconference.com.

All this reminded me of an article by Deming that I happend to see and devour while in the Math-Stat-Physics Library at UNC Chapel Hill 40 years ago. (So long ago that graduate students in those departments were given keys to this library, so we could have 24 x 7 access.)

The article: 

Deming, W. E. (1972). Code of professional conduct: A personal view. International Statistical Review, 40(2):215-219. (You can get this at http://statlab.bio5.org/foswiki/pub/Main/PapersForClassCPH685/Deming-rules.pdf.)

Deming's comments on "Fees" follow, but first ...


If you are game, as you read it, consider and respond to two questions quite pertinent to statistical consulting today. Both are based on the assumption that your client requires that you bill by the day. This may also be tied to tallying time against a retainer.

1. Client A's current project calls for you to design, develop, and test a new R function, xyz.r. This is something you will likely re-use in future projects, probably for Client A and perhaps some for other clients. Therefore, you produce something more general and more polished than if you it was merely a "one-shot" tool. xyz.r takes you 3 days to complete. If this had been a one-shot tool, the effort might would have taken 2 days. How many days do you bill/tally at this point towards Client A's project? 

2. Two months later, a project for Client B can make use of your xyz.r function. It needs no modification, so instead of taking 3.25 days of programming, you can do it in only 0.25 days. How many days do you tally towards Client B?

Here's the Deming piece. Note the phrase "my fee will be based on my subjective judgment of the relative effort expended on behalf of the engagement." How does this apply to the above questions?



V. Fees

23. The annual fee for regular consultation on a continuing basis will be agreed upon in advance, subject to change by agreement. The client may of course spread payments throughout the year at his convenience. Otherwise, he will receive a bill at the end of every year. With respect to an engagement for a single study, I will send a bill at completion, or possibly at the end of a year if completion appears to be far in the future. I do not itemize time nor regular office expenses for service: my fee will be based on my subjective judgment of the relative effort expended on behalf of the engagement. Irregular expenses such as for travel, or for tabulations done outside my office, are extra, and I may from time to time send a bill to bring them up to date.

24. I do not estimate in advance what my fee might be for participation in a single study. My fees are not competitive. I do not tailor my participation to fit a price. Another point is that the client too often has no idea about the efficiency and economy that statistical theory and techniques can contribute to his study, hence has no basis on which to compare cost against returns. Moreover, neither of us may foresee the complications and duration of a proposed study.

25. In conformity with this last paragraph and with paragraph 2, I will draw up plans for a study and for statistical participation therein only as a professional engagement. I do not make competitive proposals.

26. There will be a fee for an exploratory consultation.

27. I may, at my discretion, engage counsel of another specialist. Such an arrangement requires no clearance from the client except by specific restriction in the interest of confidentiality. Necessity for protracted assistance may call for separate financial arrangements between my client and the specialist. The choice of the specialist will ordinarily be mine.

28. I will make appropriate arrangements at my expense for competent statistical guidance during any prolonged absence that I may elect to take for vacation, meeting, lectures, or disability. If the client elects to engage some other statistician during my absence, he may do so, but at his own expense and subject to my review and satisfaction.


-------------------------------------------
Ralph O'Brien
Case Western Reserve University
-------------------------------------------

Well that bit on the fees is a thoroughly unilateral approach to things. It might be good for restricting the number of clients you actually take on (because you're so famous that you have way people beating on your door clamoring for your assistance at any price than you have time to work with), but really that strikes me as arrogant and disrespectful toward the client. I doubt very many statisticians could get most of their clients to agree to such terms.

-------------------------------------------
Steven Pierce
Associate Director
Center for Statistical Training and Consulting
Michigan State University
-------------------------------------------






Original Message:
Sent: 06-11-2012 17:16
From: Ralph O'Brien
Subject: Deming on consulting fees


This relates to two ASA-related eGroup postings I've seen recently:

(1) the discussion within the Consulting Section eGroup about working under retainer arrangements.

(2) the announcement of the next (68th) Annual Deming Conference on Applied Statistics, December 2-7, 2012; Atlantic City, NJ. The full program is available at www.demingconference.com.

All this reminded me of an article by Deming that I happend to see and devour while in the Math-Stat-Physics Library at UNC Chapel Hill 40 years ago. (So long ago that graduate students in those departments were given keys to this library, so we could have 24 x 7 access.)

The article: 

Deming, W. E. (1972). Code of professional conduct: A personal view. International Statistical Review, 40(2):215-219. (You can get this at http://statlab.bio5.org/foswiki/pub/Main/PapersForClassCPH685/Deming-rules.pdf.)

Deming's comments on "Fees" follow, but first ...


If you are game, as you read it, consider and respond to two questions quite pertinent to statistical consulting today. Both are based on the assumption that your client requires that you bill by the day. This may also be tied to tallying time against a retainer.

1. Client A's current project calls for you to design, develop, and test a new R function, xyz.r. This is something you will likely re-use in future projects, probably for Client A and perhaps some for other clients. Therefore, you produce something more general and more polished than if you it was merely a "one-shot" tool. xyz.r takes you 3 days to complete. If this had been a one-shot tool, the effort might would have taken 2 days. How many days do you bill/tally at this point towards Client A's project? 

2. Two months later, a project for Client B can make use of your xyz.r function. It needs no modification, so instead of taking 3.25 days of programming, you can do it in only 0.25 days. How many days do you tally towards Client B?

Here's the Deming piece. Note the phrase "my fee will be based on my subjective judgment of the relative effort expended on behalf of the engagement." How does this apply to the above questions?



V. Fees

23. The annual fee for regular consultation on a continuing basis will be agreed upon in advance, subject to change by agreement. The client may of course spread payments throughout the year at his convenience. Otherwise, he will receive a bill at the end of every year. With respect to an engagement for a single study, I will send a bill at completion, or possibly at the end of a year if completion appears to be far in the future. I do not itemize time nor regular office expenses for service: my fee will be based on my subjective judgment of the relative effort expended on behalf of the engagement. Irregular expenses such as for travel, or for tabulations done outside my office, are extra, and I may from time to time send a bill to bring them up to date.

24. I do not estimate in advance what my fee might be for participation in a single study. My fees are not competitive. I do not tailor my participation to fit a price. Another point is that the client too often has no idea about the efficiency and economy that statistical theory and techniques can contribute to his study, hence has no basis on which to compare cost against returns. Moreover, neither of us may foresee the complications and duration of a proposed study.

25. In conformity with this last paragraph and with paragraph 2, I will draw up plans for a study and for statistical participation therein only as a professional engagement. I do not make competitive proposals.

26. There will be a fee for an exploratory consultation.

27. I may, at my discretion, engage counsel of another specialist. Such an arrangement requires no clearance from the client except by specific restriction in the interest of confidentiality. Necessity for protracted assistance may call for separate financial arrangements between my client and the specialist. The choice of the specialist will ordinarily be mine.

28. I will make appropriate arrangements at my expense for competent statistical guidance during any prolonged absence that I may elect to take for vacation, meeting, lectures, or disability. If the client elects to engage some other statistician during my absence, he may do so, but at his own expense and subject to my review and satisfaction.


-------------------------------------------
Ralph O'Brien
Case Western Reserve University
-------------------------------------------

Deming's approach is too loose for my style but I would not go so far as to call it "wrong".  The important thing is to come to an understanding with the potential client up front and to honestly negotiate later changes if and when they seem appropriate.

When thinking about the Client A/Client B scenario we should bear in mind that we learn new things with most consulting assignments.  We can therefore argue that Client A benefited from whatever we did for some former client just as Client B benefits from what we did for Client A.

I just charge for whatever time I spend on the client's project.  If learning from previous projects makes me more valuable then I can raise my hour rate.  In the scenario you described, discounting the charges for Client A in the expectation of a possible Client B and charging Client B for something I developed for Client A would lead to unworkable complexity and invoices that would raise clients eyebrows.

PS unrelated to this particular post.  I deleted the original message because including the original leads to very long postings arranged in reverse chronological order and because everyone can follow the original messge in the message tree.  Does that make sense to others?  Do we want to encourage or discourage that as a standard practice for this discussion group.
-------------------------------------------
Emil M Friedman, PhD
emil.friedman@alum.mit.edu (forwards to day job)
emilfrie@alumni.princeton.edu (home)
http://www.statisticalconsulting.org
-------------------------------------------

Deming's approach may work for a world famous genius. I doubt it would work for most and I know it would not work for me.

The only part I agree with is that I do not set overall estimates. But I bill in small increments.

I also almost never have such long term contracts.  Many of my jobs take less than 10 hours of my time.

-------------------------------------------
Peter Flom
-------------------------------------------






Original Message:
Sent: 06-12-2012 10:00
From: Emil Friedman
Subject: Deming on consulting fees

Deming's approach is too loose for my style but I would not go so far as to call it "wrong".  The important thing is to come to an understanding with the potential client up front and to honestly negotiate later changes if and when they seem appropriate.

When thinking about the Client A/Client B scenario we should bear in mind that we learn new things with most consulting assignments.  We can therefore argue that Client A benefited from whatever we did for some former client just as Client B benefits from what we did for Client A.

I just charge for whatever time I spend on the client's project.  If learning from previous projects makes me more valuable then I can raise my hour rate.  In the scenario you described, discounting the charges for Client A in the expectation of a possible Client B and charging Client B for something I developed for Client A would lead to unworkable complexity and invoices that would raise clients eyebrows.

PS unrelated to this particular post.  I deleted the original message because including the original leads to very long postings arranged in reverse chronological order and because everyone can follow the original messge in the message tree.  Does that make sense to others?  Do we want to encourage or discourage that as a standard practice for this discussion group.
-------------------------------------------
Emil M Friedman, PhD
emil.friedman@alum.mit.edu (forwards to day job)
emilfrie@alumni.princeton.edu (home)
http://www.statisticalconsulting.org
-------------------------------------------

Good Morning Ralph,

How's it going? Great question. In our kind of work, we run into this though often on a bigger scale. Here are my thoughts.

Client A has asked me to do something and that's what I should bill them for. Whether I do all the work you described at once or polish it after I deliver to Client A, I would only bill them for the work they asked me to do. The polishing would be on my own time.

For future clients it's different. The more general question is if you make an investment that reduces the amount of time it takes for a consulting service, how do you bill the client going forward? If you take a class that shortens your analysis time, get a better software package for data cleansing, buy a faster computer, etc, do you bill the old number of hours or the new number of hours that it takes you? I see it as the same question.

The other aspect, of course, is your rate. If you invest more money into your skills, then (we hope!) your services become more valuable per hour and your rate can be greater. The net effect might be cheaper for the client and more for you. (Unless your fee for new clients is somehow fixed.)

I think that this approach relates to Deming's quote well, since his fee is a combination of hours and rate, which we might think of as the quantity and quality of his relative effort.

Just my two cents.

-------------------------------------------
Charles Kincaid
Engagement Director
Experis Business Intelligence and Analytics
chuck.kincaid@experis.com

Webmaster
Section on Statistical Consulting
-------------------------------------------


Original Message (with bits deleted):
Sent: 06-11-2012 17:16
From: Ralph O'Brien
Subject: Deming on consulting fees

If you are game, as you read it, consider and respond to two questions quite pertinent to statistical consulting today. Both are based on the assumption that your client requires that you bill by the day. This may also be tied to tallying time against a retainer.

1. Client A's current project calls for you to design, develop, and test a new R function, xyz.r. This is something you will likely re-use in future projects, probably for Client A and perhaps some for other clients. Therefore, you produce something more general and more polished than if you it was merely a "one-shot" tool. xyz.r takes you 3 days to complete. If this had been a one-shot tool, the effort might would have taken 2 days. How many days do you bill/tally at this point towards Client A's project? 

2. Two months later, a project for Client B can make use of your xyz.r function. It needs no modification, so instead of taking 3.25 days of programming, you can do it in only 0.25 days. How many days do you tally towards Client B?

Here's the Deming piece. Note the phrase "my fee will be based on my subjective judgment of the relative effort expended on behalf of the engagement." How does this apply to the above questions?

This one seems pretty straight forward.  Your dirst client gets a huge benefit from your general program, as you will document it completely.  Ralph, you are especially good at that and your client benefits from the QA that goes into developing the general macro.  So cliewnt A must pay for this in full.

There is no reason to charge client B more than the actual work you put in. 

My thoughts,

Best wishes,


Jon

-------------------------------------------
Jon Shuster
University of Florida
-------------------------------------------






Original Message:
Sent: 06-12-2012 11:28
From: Charles Kincaid
Subject: Deming on consulting fees

Good Morning Ralph,

How's it going? Great question. In our kind of work, we run into this though often on a bigger scale. Here are my thoughts.

Client A has asked me to do something and that's what I should bill them for. Whether I do all the work you described at once or polish it after I deliver to Client A, I would only bill them for the work they asked me to do. The polishing would be on my own time.

For future clients it's different. The more general question is if you make an investment that reduces the amount of time it takes for a consulting service, how do you bill the client going forward? If you take a class that shortens your analysis time, get a better software package for data cleansing, buy a faster computer, etc, do you bill the old number of hours or the new number of hours that it takes you? I see it as the same question.

The other aspect, of course, is your rate. If you invest more money into your skills, then (we hope!) your services become more valuable per hour and your rate can be greater. The net effect might be cheaper for the client and more for you. (Unless your fee for new clients is somehow fixed.)

I think that this approach relates to Deming's quote well, since his fee is a combination of hours and rate, which we might think of as the quantity and quality of his relative effort.

Just my two cents.

-------------------------------------------
Charles Kincaid
Engagement Director
Experis Business Intelligence and Analytics
chuck.kincaid@experis.com

Webmaster
Section on Statistical Consulting
-------------------------------------------


Original Message (with bits deleted):
Sent: 06-11-2012 17:16
From: Ralph O'Brien
Subject: Deming on consulting fees

If you are game, as you read it, consider and respond to two questions quite pertinent to statistical consulting today. Both are based on the assumption that your client requires that you bill by the day. This may also be tied to tallying time against a retainer.

1. Client A's current project calls for you to design, develop, and test a new R function, xyz.r. This is something you will likely re-use in future projects, probably for Client A and perhaps some for other clients. Therefore, you produce something more general and more polished than if you it was merely a "one-shot" tool. xyz.r takes you 3 days to complete. If this had been a one-shot tool, the effort might would have taken 2 days. How many days do you bill/tally at this point towards Client A's project? 

2. Two months later, a project for Client B can make use of your xyz.r function. It needs no modification, so instead of taking 3.25 days of programming, you can do it in only 0.25 days. How many days do you tally towards Client B?

Here's the Deming piece. Note the phrase "my fee will be based on my subjective judgment of the relative effort expended on behalf of the engagement." How does this apply to the above questions?

I agree that improvements in one's professional skills achieved through learning and/or developing better statistical methods should be reflected by an appropriate increase in the statistician's future hourly consulting fee, instead of being charged as additional time spent on any particular project.  

-------------------------------------------
Edith Zang
Independent Consultant
NYCASA
-------------------------------------------






Original Message:
Sent: 06-12-2012 11:28
From: Charles Kincaid
Subject: Deming on consulting fees

Good Morning Ralph,

How's it going? Great question. In our kind of work, we run into this though often on a bigger scale. Here are my thoughts.

Client A has asked me to do something and that's what I should bill them for. Whether I do all the work you described at once or polish it after I deliver to Client A, I would only bill them for the work they asked me to do. The polishing would be on my own time.

For future clients it's different. The more general question is if you make an investment that reduces the amount of time it takes for a consulting service, how do you bill the client going forward? If you take a class that shortens your analysis time, get a better software package for data cleansing, buy a faster computer, etc, do you bill the old number of hours or the new number of hours that it takes you? I see it as the same question.

The other aspect, of course, is your rate. If you invest more money into your skills, then (we hope!) your services become more valuable per hour and your rate can be greater. The net effect might be cheaper for the client and more for you. (Unless your fee for new clients is somehow fixed.)

I think that this approach relates to Deming's quote well, since his fee is a combination of hours and rate, which we might think of as the quantity and quality of his relative effort.

Just my two cents.

-------------------------------------------
Charles Kincaid
Engagement Director
Experis Business Intelligence and Analytics
chuck.kincaid@experis.com

Webmaster
Section on Statistical Consulting
-------------------------------------------


Original Message (with bits deleted):
Sent: 06-11-2012 17:16
From: Ralph O'Brien
Subject: Deming on consulting fees

If you are game, as you read it, consider and respond to two questions quite pertinent to statistical consulting today. Both are based on the assumption that your client requires that you bill by the day. This may also be tied to tallying time against a retainer.

1. Client A's current project calls for you to design, develop, and test a new R function, xyz.r. This is something you will likely re-use in future projects, probably for Client A and perhaps some for other clients. Therefore, you produce something more general and more polished than if you it was merely a "one-shot" tool. xyz.r takes you 3 days to complete. If this had been a one-shot tool, the effort might would have taken 2 days. How many days do you bill/tally at this point towards Client A's project? 

2. Two months later, a project for Client B can make use of your xyz.r function. It needs no modification, so instead of taking 3.25 days of programming, you can do it in only 0.25 days. How many days do you tally towards Client B?

Here's the Deming piece. Note the phrase "my fee will be based on my subjective judgment of the relative effort expended on behalf of the engagement." How does this apply to the above questions?