Discussion: View Thread

Hi everyone,

I just finished reading an interesting article in the latest issue of AMSTATNEWS (Communicating with Clients, by Robert Starbuck).

The article advocates that, as consultants, we should clarify with the client what will represent 'success'. 

In practice, we never clarify with the client what will represent 'failure', though 'failure' of a project can unfortunately be a very real possibility.  'Failure' can encompass a variety of aspects, including failure to meet deadlines, failure to understand and meet client expectations, failure to communicate consultant availability, needs and expectations, failure to operate within the allowed budget or to provide a realistic quote for consulting services, failure to communicate effectively with the client given their preferred style of communication, failure to use appropriate statistical tools to tackle a problem, failure to present and explain adequately the meaning and implications of statistical results, and the list goes on.

I am not suggesting that we should tackle the issue of 'failure' upfront with our clients, but I am certain that many of us have built-in alarm flags which help us detect when the project is veering in a dangerous direction and allow us to make course corrections (if that is at all possible). 

It would be useful, in my view, for us as a group to have a conversation about 'failure' and how we can avoid it in the context of statistical consulting.  Indeed, 'failure' can be very costly for consultants, especially if it leads to litigation, refusal to pay consulting services or loss of future business.  Another, more insidious aspect of 'failure' is the personal toll it can take on everyone involved (not just the consultant, but also the clients).  While it is not as glamorous as 'success', 'failure' is the other side of the coin - a reminder that, if things have the potential to go wrong, they will go wrong, unless we step in and prevent their derailment.

Are there any "best practices" for mitigating and/or managing 'failure' that we should all be aware of in our field?  Is there any collective wisdom that we could all build on in order to avoid 'failure'?  Can we develop some level of personal/business protection against 'failure' in our consulting practices? 

I would be curious to hear your thoughts and inputs on these issues. 

Thanks!

Isabella

-------------------------------------------
Isabella Ghement
Ghement Statistical Consulting Company Ltd.
-------------------------------------------

Hi Isabella,

As part of the sales process, it's great to ask if the client has worked with statistical consultants before and if so, what worked well and what didn't.  If you ask them to talk a bit more about what didn't work well and what they would have wanted to be different, you can learn very specifically what are "hot buttons" for the client and how specifically to avoid them.  At the same time, if you ask them to talk more about what did work well, you'll understand what is important to them.  If they haven't worked with a consultant before you can frame the question a little differently and still get the same information. 

The second practice is to consistently write down after each interaction what went well and what you would like to do differently next time.  Then review your list before you go into the next interaction. 

Hope this finds you well!

-------------------------------------------
Colleen Mangeot
Biostatistician
Cincinnati Children's Hospital
-------------------------------------------

This is indeed an interesting question and a topic that all statisticians should be thinking about.  To answer the question "is failure even an option?", the simple answer is "of course it is".  But the REAL issue is not about failure, but about defining what constitutes a failure and then managing the risk of failure.  Whether or not this discussion happens literally between a client and a consultant, the understanding needs to be in place.

For example, failure of a study to show statistical significance is not necessarily a failure of the consultant or the client.  It could be due to inadequate assumptions, poor study conduct, incorrect choice of endpoints, and so on.  However, I consider a study that fails to demonstrate significance to be am important part of the knowledge going forward.  If the information gained can help minimize or eliminate the risks for the next study then it is has not failed.  On the flip-side, as consultants, we are often "pushed" to come up with assumptions that will help us to estimate smaller sample size, or take on riskier design methods.  If we allow this, knowing that the risks of failure are now higher, AND we do NOT express our concerns for these risks, then we are also at fault for the failure.  As I like to remind my clients when I have to give them information they do not want to hear, they are paying me for my experiences, expertise and opinions.  If they are not willing to utilize the knowledge and experience I bring to the table, that is not a failure on my part, but a CHOICE they get to make as the client.

How do I minimize failure as a statistical consultant: 1) READ and interact with colleagues; 2) stay up to date to current trends, methods, and pitfalls; 3) explain as clearly as possible what you see and the risks and benefits of any recommendations you make. 4) be willing to admit when you don't understand a concept or when you need more details from the client. 5) keep the client as informed as you can, without overburdening them with details.

-------------------------------------------
Susan Spruill
Statistical Consultant
-------------------------------------------







Original Message:
Sent: 11-27-2013 12:17
From: Colleen Mangeot
Subject: Is failure even an option in statistical consulting?

Hi Isabella,

As part of the sales process, it's great to ask if the client has worked with statistical consultants before and if so, what worked well and what didn't.  If you ask them to talk a bit more about what didn't work well and what they would have wanted to be different, you can learn very specifically what are "hot buttons" for the client and how specifically to avoid them.  At the same time, if you ask them to talk more about what did work well, you'll understand what is important to them.  If they haven't worked with a consultant before you can frame the question a little differently and still get the same information. 

The second practice is to consistently write down after each interaction what went well and what you would like to do differently next time.  Then review your list before you go into the next interaction. 

Hope this finds you well!

-------------------------------------------
Colleen Mangeot
Biostatistician
Cincinnati Children's Hospital
-------------------------------------------

Appropriately defining the objective is often critical to the definition of success or failure.  The blog you refer to (http://stattrak.amstat.org/2013/11/01/communicating-with-clients/) mentions an example where the objective is to "increase the output of a production line".  Perhaps it is possible to do so without sacrificing quality; perhaps not.  A more achievable objective might be, "Determine whether and by how much production can be increased and at what cost."  

Years back a manager at a former employer said, "Emil, I heard that the XXX project failed."  I told him, "No, it did not fail.  We were able to show that changing to the supposedly better (and more expensive) bl compound would not lengthen bladder life.  Therefore shelving the project saved us a lot of money." 

Another time a lawyer wanted me to explain the statistics behind some evidence that the prosecution would be presenting.  I told him up front that that I would be charging a hourly rate and that there was no way to know whether the explanation would help or hurt his client until we saw what the evidence.

It's a lot better and a lot more honest to explicitly agree up front that the objective is "Determine whether" and not "Prove that".  If the client is uncomfortable with that, it's important to then discuss whether it's worth going ahead with the project and what will happen if it can't be done.

-------------------------------------------
Emil M Friedman, PhD
emil.friedman@alum.mit.edu (forwards to day job)
emilfrie@alumni.princeton.edu (home)
http://www.statisticalconsulting.org
-------------------------------------------

Isabella,

Interestingly enough, today I received an email about an upcoming webinar that included "defining failure" as a topic.  Here's a portion of that email (with my bold type added to highlight the pertinent topic):

Featured Topics 

 - Examining the shifting global economic trends that compel outsourcing of highly-skilled, technical functions
 - Exploring the top 5 reasons most outsourced initiatives disappoint
 - Establishing contingency plans for when outsourcing stops making sense
 - Analyzing how sponsor-provider partnerships succeed by prioritizing the strategic imperatives beyond cost savings
 - Why both parties agreeing on the definition of "failure" is key to success
 - An in-depth case study of the compelling reasons to continue outsourcing: gaining project flexibility & expertise lacking in-house while keeping sponsor focus on core competencies 

For context, this webinar is about outsourcing support such as biostatistics, statistical programming, and data management for clinical trials.  (I am not in any way associated with this free event on Dec 4th.  It is sponsored by DIA and Cytel.)
http://www.diahome.org/en-US/Meetings-and-Training/Find-Meetings-and-Training/Meeting-Details.aspx?ProductID=2683310&EventType=Webinar

I share this just to let you know you aren't the only person concerned about agreeing up-front with a client on what "failure" is.

-------------------------------------------
Mark Martin
Siemens Healthcare Diagnostics
-------------------------------------------