Tuesday, March 25, 2008

What's a ScrumMaster Worth?

You may have noticed that some people are feeling a recession out there. So money can be a bit tighter.

So, can you afford a good ScrumMaster?

The answer is obviously yes, and in fact they are even in greater need (since there is greater urgency).

Now, let's unwrap this from a financial viewpoint. We will use a simple example, and provide a simple spreadsheet. Hopefully you can take these ideas, use your own local numbers, and have a good conversation so the right thing happens.

A caveat: This post is not suggesting that the ScrumMaster is the end-all and be-all. We are asserting that the ScrumMaster can have a big influence on the productivity of a team. And that better ScrumMasters can have a lot more influence. And that the best ScrumMasters are rare.

* * *

Imagine a team of 8 that costs $1,000,000 per year. Including the Product Owner and ScrumMaster.

That team produces Business Value at some multiple of its cost. Let's take the case that that multiple is 7x. So the team produces $7,000,000 in BV per year. (One can think of BV being NPV (net present value) but it could be measured, originally at least, many other ways.)

Let's take the case that the team has an "ok" ScrumMaster, but the team is increasing velocity at only 10% per year. (Assume at the beginning of the year they are running at 150% of waterfall velocity.) Assume the "ok" ScrumMaster is making, all-in, $125K.

Now assume a "better" ScrumMaster who can double the productivity of the team in one year. (He does this by removing impediments or having them removed.) And assume that the better ScrumMaster (if he is available) costs $250K per year.

(NB. I hope you are noting that the numbers we are using are simple, round and convenient. You have to identify your own numbers. Nonetheless, I will call the numbers in this post, hopefully, inspirational.)

My, my, my. Very expensive dude, isn't he.

Should we invest in the better ScrumMaster?

Well, let's look at this some more. We assume that the Product Owner has or can discover new Product Backlog items so that the average BV of stories will not decline over the year. And let's assume the better ScrumMaster does not improve productivity by helping her (the Product Owner) discover more business value. Let's further assume the better SM improves (enables the team to improve) velocity roughly equally over the year. So that, over the next year the team will produce $10.5 million in business value (rather than $14 million, if the jump were to happen immediately).

I have also assumed the situation (the team, the managers, the company, etc) will allow the better SM to help enable the team to reach a 2x level.

(N.B. The conversation is about value in relation to cost. It is not, and never was, purely a cost consideration.)

So, for an added investment of $125K, the firm will get an extra $3.15 million. Is this a good business decision?

Well, we don't quite know yet.

Let's assume the better SM finds impediments that cost $1 million to fix (training, SW, HW, etc), so that, in simple terms, the firm must invest a total of $1.125 million total to get $3.15 million.

What do you think? Is this a good business decision in a recession?

One could discuss my assumptions endlessly. Discuss a little if you must; then do an experiment where you are. We are all interested in your results.

(To update this algorithm with your own assumptions, download the XLS file here. And revise it.)

N.B. Scrum was built to help teams get 5x to 10x productivity gains. I think a 2x gain in one year is conservative (low, easily reached in a normal situation). So, there is more juice in the orange. There is also the possibility that you have a "dead core" and further productivity improvements are not possible. More on this in later posts.
N.B. A key principle of Agile is sustainable pace. So in Scrum, the gains are not made by driving the team through a "death march". Unfortunately, this still has to be said for some people.

15 comments:

Anonymous said...

"but the team is not increasing velocity" - seems not absolutely correct.

Any scrum master will provide constant team productivity at the end and velocity won't be increasing.

Time to change SM again? ;)

And what is "ok", how to judge?

Joe Little said...

Yes. So my assumption with the "OK" SM is that he has gotten them to 150% and stopped. (You might call this a "bad" SM.)

Every SM should be removing impediments so that team velocity can increase. The only difference should be that better SMs are doing this "faster" (more effectively).

I will re-read my original entry and may revise it.

Thanks.

Peter said...

Hi Joe,

Thought provoking analysis. But I think you leave out the risk analysis: How do I know that the new scrum master will bring this value? Sounds like a job for Value (and Successs) based pricing.

Cheers,

Peter

Joe Little said...

Hi Peter,

Yes, we did not talk about risk explicitly. My general experience is that risk is best addressed by just doing what appears to bring in the highest return. 300% in one year usually seems high enough to me.

We also did not talk about the longer-term effects of the change (outside one year, outside the one team).

I have also seen people get sooo absorbed in risk, that they would not try anything (new). Change is too much risk for them; when in fact staying the same is actually too much risk.

Also I wanted to keep it simple. A lot of people won't "get it" if you make it too complex.

So, how would you bring in some risk analysis? Specifically. (To me, the door is open when one says "assume". And that's also why I built the spreadsheet for others to try their own assumptions.)

Thanks, Joe

Peter said...

Perhaps I should turn this one around? How do you know that your Scrummaster is bad? How do you know the new one will be better?

An essential part of Scrum is continuous improvement. So if you change scrummasters, the new one should be better than the old one, just because after 6 months, the improvements brought to light by the team should have an impressive cumulative effect.

Does this mean the new scrummaster is better than the old one? not necessarily, because time works on the side of the new guy.

Anonymous said...

This seems slightly flawed. You are saying a good SM can increase the efficiency. How exactly is efficiency measured more features? So if the team produces more features they are more efficient?

So how does this scenario fit in. Any team following Scrum may not increase efficiency but may increase the quality and documentation of their output but still produce the same number of features.

I work with a team who has followed scrum for 3 years in that time this team has:
- increased quality of the product dramatically
- produced excellent user documentation
- less stressed at the end of a release cycle
- more likely to release a product that is fit for purpose
- more likely to deliver the features promised at the beginning of a cycle.
- has even once delivered a large feature extra to requirements.

So tell me is that a successful or unsuccessful SM.

Also I thought one of the principles of scrum was not to try increase velocity. Velocity is what it is, a measure of constant output of a team.

George Dinwiddie said...

Joe, you say: Now assume a "better" ScrumMaster who can double the productivity of the team in one year. (He does this by removing impediments or having them removed.)

This seems like quite an assumption to me. And a lot of impediments left in the organization, such that all the attention goes there and none to improving technical practices.

It also violates Jerry Weinberg's Ten Percent Promise Law (from The Secrets of Consulting). I suppose that's part of why it doesn't seem like a believable assumption.

Where did you get the numbers you're using?

Joe Little said...

Hi Peter,

You are right to point out that few things are known with absolute certainly. Let us say the difference between the two SMs seems obvious to the observer. And so it has appeared to me. Afterwards, sometimes I was right and sometimes I was too conservative. And sometimes I was wrong (ie, the "better" person was not actually better).

But we have to make business decisions with uncertainty.

You could argue that time (with the team) is on the side of the old guy. That new/old thing is only one of the root causes of success (or lack of success).

But let's get back to the main issue I was trying to address. If you are a manager, you must decide what to pay a SM. How do you decide what's reasonable? I am proposing that it be based on the net effect on BV delivered.

Regards, Joe

Joe Little said...

Hi Anon.,

How should effectiveness (I somewhat prefer this term) be measured? Great question. I certainly did not deal with it directly. Beyond the scope of this small post.

My short answer would be "velocity", which I would normally do as story points done in each iteration. And I assumed that stories were of relatively constant BV (on average).

One could argue that a better metric is BV delivered per iteration. That is a little harder to explain in one short post. And, in my experience, typically harder to do in practice.

I think one of the principles of Scrum is EVERYONE wants to help the team increase velocity. Not by working harder, but by removing (reducing) impediments.

What your team did sounds very good. That may have meant that a lot more BV was actually delivered. (I am, as you may be, a little cynical about any metric. The real thing is people, and how what we do affects them.)

We continually strive for perfection, yet never achieve it. We always look back and ask "how much more could we have achieved if...?" And then we take action in the present.

Thanks, Joe

Joe Little said...

Hi George,

Honored that you are reading the blog.

Yes, it was an assumption. I have seen double happen in a bunch of cases. And more. And I have talked with enough people who have gotten more that I am embarrassed to say I can only produce (most likely outcome) double.

I have also had cases myself where the team or the firm or others would not allow double to happen. Very frustrating. I assumed this problem away in my post.

I would agree with your implication that technical practices are normally very important impediments. In each situation, deciding on and acting on the most important impediments is difficult. (And not acting on lower priority "obvious" impediments is also hard.) I do not wish to minimize the thought and work involved.

Obviously, in this post I did not feel constrained by Jerry Weinberg's law. I like his stuff (as much of it as I have read).

Anyway, I am more concerned about the algorithm than the specific numbers in it. Unless the numbers are so low that the SM is not paid well. In which case I would say that someone has insufficient imagination to believe he/she/"we" can have a better life.

In general, I would agree with the adage: Under promise, over deliver.

What do your typical numbers say?

Thanks, Joe

Joe Little said...

To all,

So far I am pretty happy with the algorithm. We can argue about my assumptions (or yours) all day. To me, the specific numbers are not nearly as important as the thought process (and that the value of a good SM is more obvious to more people).

I would say the algorithm is a bit simplistic, but generally good enough for most decision-making in this case (do we hire him? and for how much?).

I welcome additional comments.

Regards, Joe

linda.cook2 said...

Nice, juicy,controversial topic! You could make a similar assertion for all SCRUM roles, the PO and the Dev's alike. But I won't.

The greater question is should you change out a SM for a higher paid SM, who supposedly has better skills and will therefore be able to raise the velocity of the team? The idea of this gives me chills - the scary kind. Let's start with who make the decision to pull the plug on a SM. Senior leaders in an Agile organization should be facilitating a learning environment in the hopes to grow their Scrum teams. When you examine the costs of hiring a new SM, let's say $15K real money, one might easily decide they can 'afford' a better SM. Certainly an additional $15k cost won't hurt your model.

What does it cost an organization, presumably in transition to Scrum, (less than 2 yrs) to remove folks from a Scrum team, in favor of a higher paid SM who should be able to more productivity from the team?
Now you have entered the real danger zone! It is one thing to remove someone from an organization for non-performance. When a non-performing SM is hurting the team, corrective action is needed quickly. People are not so fungible that you can arbitrarily move out one player and move in another without impacting the team. This can be even more critical when you are talking about the SM role. The span of influence the SM has in the organization is often greater than many of the team members. A change in this role can have a greater impact on the organization.

IMHO, an org's money is better spent developing the SM to help them become more effective in their role (such as better at removing obstacles). The practice of developing existing staff members can be overlooked when companies are facing tough challenges. Getting another higher-paid person in the role is not a guarantee of success. I'd rather spend money developing folks internally and then rewarding them financially for higher performance.

My banter has gone long enough, so I'll close with these thoughts. You get what you pay for and a good SM is worth their weight in gold.

Thanks for the juicy topic!
Linda

Joe Little said...

Hi Linda,

Honored that you are reading the blog.

Well, probably the "ok" vs "better" SM case I set up was not the best choice. It is not really what I am getting at. And changing people in mid-stream is certainly not the key issue for me. I see your points about that, having had to deal with or observe similar situations.

Yes, agree with your point that the same could be said about the value of other players. I set up this scenario to isolate one player (the SM).

I do find that the value of other players is more readily seen by more people. This is in part because of the nature of the role (and the observers).

I did not mean to imply that the new SM would be internal or external. He/she might be an internal transfer. Again, in any case, the switching has other effects, and these need to be considered, as you suggest.

Regards, Joe

George Dinwiddie said...

Joe,

I don't have any numbers to report. For myself, it's generally enough to answer, are things better, worse, or about the same. If you keep doing that, over various time spans, it seems to push for a general improvement.

Productivity is a devilish thing to measure. And I'm not sure the productivity of a team is up to the scrummaster. I think that a coach might be a better choice for improving productivity. Of course, an excellent scrummaster may also be a coach, and maybe that's what you're really talking about.

My experience is that organizations take project managers, architects, and technical leads and anoint them as scrummasters. Some do quite well. Others are terrible in their new role.

There's a whole lot of stuff involved in helping a team. Some is technical, some interpersonal. I would be sceptical of any numerical predictions of improvement. And that makes predicting the appropriate pay somewhat problematical.

It seems to me that increases in pay are usually after the fact, when the value has been demonstrated. And were I in a position to change from a run-of-the-mill scrummaster in hand, to a rockstar asking for a job, I might well choose to invest in the one I have. I could fund some education and, if it pays off, then upgrade the salary. Seems less risky, to me.

BTW, someone trying this approach might find the AYE Conference (http://ayeconference.com/) to be a good educational tool. I say this as a satisfied customer. I have not financial interest in the conference (though some of my friends do).

Joe Little said...

Hi George,

I like your observations.

I do think for lots of people, you need numbers. Yes, I would agree that you have to always ask "do we really believe the numbers". The numbers are high level, like velocity, not individual.

I don't see productivity as devilish. It is not always totally obvious each iteration, but in the end, it should be obvious.

Yes, I agree SMs are quite variable. And I think I would agree that what you call an Agile coach is probably what I think every SM should be. (Most now, some to soon become. Maybe this is a bit aggressive, given how new Agile is, how fast it is growing.)

Still, my experience is that there are a lot of "OK" ScrumMasters out there. (Many reasons for this.) So, getting the team in position to double productivity is not so hard for a good SM.

Yes, I am also assuming a team that can double and a management team that will support. If they understand well, this is fairly common I think. But certainly not always.

You noted where a lot of SMs come from. My experience is that people don't understand this "odd" position well; so often we have the wrong people in that position. And where we have the right people, we often haven't explained the job well.

My experience is that there are sooo many impediments. We just have to identify and fix the biggest ones (you rightly mention that there are lots of types of impediments).

Yes, you don't bring in an expensive player just because he says so. But one does, one should and one must make forward looking decisions. And this includes bringing in special talent like agile coaches or whatever.

Again, my post is really about the logic for why a good SM is valuable. The "story" around that is only incidental (an example).

Should you invest in internal people? Of course. And I have always found that external people have a place too (as consultants or as hire-ins). Needed if you want to change the culture.

Thanks, Joe