Paul Goodman [email protected] Introducing Software Metrics via Benchmarking.

22
Paul Goodman [email protected] Introducing Software Metrics via Benchmarking

Transcript of Paul Goodman [email protected] Introducing Software Metrics via Benchmarking.

Page 1: Paul Goodman Paul.Goodman@metagroup.com Introducing Software Metrics via Benchmarking.

Paul Goodman

[email protected]

Introducing Software Metrics via Benchmarking

Introducing Software Metrics via Benchmarking

Page 2: Paul Goodman Paul.Goodman@metagroup.com Introducing Software Metrics via Benchmarking.

2© 2004 META Group Consulting 01252 819494 metagroup.com

Topics for Discussion

Metrics Programme risks and reasons for failure

Benchmarking as a means to mitigating these risks

Moving beyond Benchmarking

Page 3: Paul Goodman Paul.Goodman@metagroup.com Introducing Software Metrics via Benchmarking.

3© 2004 META Group Consulting 01252 819494 metagroup.com

“80% of software measurement initiatives fail” (Howard Rubin – META Group)

Failure is defined as: (within 2 years);

Initiative is formally cancelled!

Resources are quietly reassigned

The work continues in some way but the results are ignored within the organisation

Introducing Metrics – “A Series of Unfortunate Events” (Lemony Snickett)

Introducing a metrics programme to an organisation is a risky business

Page 4: Paul Goodman Paul.Goodman@metagroup.com Introducing Software Metrics via Benchmarking.

4© 2004 META Group Consulting 01252 819494 metagroup.com

Key Reasons for Metrics Programme Failure

Expectations are not managed! Data becomes confused with information – of

which there is a serious lack! We build an ivory tower – we don’t involve the

senior or project managers (the customers of the measurement initiative)!

The devil is in the detail – or how many Function Points can we get on the head of a pin!

Senior management commitment is neither gained nor maintained!

Page 5: Paul Goodman Paul.Goodman@metagroup.com Introducing Software Metrics via Benchmarking.

5© 2004 META Group Consulting 01252 819494 metagroup.com

Some Good News

Over the last four years, I have seen more successful measurement initiatives than at any other period (and I have been in this game for a long time)

These have included measurement initiatives in: Financial organisations Local and Central Government sector Telecoms Manufacturing sector

Page 6: Paul Goodman Paul.Goodman@metagroup.com Introducing Software Metrics via Benchmarking.

6© 2004 META Group Consulting 01252 819494 metagroup.com

Why the Change?

My observation is that there are two main factors contributing to this greater success: More narrow focus – Don’t try to do too much,

too quickly The client side “pull” for benchmarking in its

various forms, primarily:– Traditional comparative benchmarking– Value for money and viability assessments of

proposals– Outsource contract management

As an industry we could be approaching a 50/50 measurement initiative success rate (so it is

still risky!)

Page 7: Paul Goodman Paul.Goodman@metagroup.com Introducing Software Metrics via Benchmarking.

7© 2004 META Group Consulting 01252 819494 metagroup.com

Measurement via Benchmarking

What follows is not just a sales pitch for Meta Group Benchmarking – honest!

Benchmarking brings its own issues and pains but also some significant benefits

First point to note: Benchmarking is based on the provision of

metrics data I want to consider the move to the point where

that data collection becomes inherently beneficial to the organisations providing the data

Page 8: Paul Goodman Paul.Goodman@metagroup.com Introducing Software Metrics via Benchmarking.

Can Benchmarking Address the Causes of Failure? (Reverse order)

Can Benchmarking Address the Causes of Failure? (Reverse order)

Page 9: Paul Goodman Paul.Goodman@metagroup.com Introducing Software Metrics via Benchmarking.

9© 2004 META Group Consulting 01252 819494 metagroup.com

Management Commitment

Facts of Life: Benchmarking costs money!

(ISBSG is a much cheaper alternative but limited in scope)

Spending money involves management signoff – often at a very senior level

Involving a third party (the “Benchmarker”) has political implications (Non-disclosure agreements at the very least)

First delivery of results is always mandated to be within months – not years! I.e. within the “management window” of 2 years

Page 10: Paul Goodman Paul.Goodman@metagroup.com Introducing Software Metrics via Benchmarking.

10© 2004 META Group Consulting 01252 819494 metagroup.com

Management Commitment

All of this generates: Visibility of the benchmarking initiative within

the organisation (Senior) management involvement during the

benchmark procurement Focus on results rather than process (not

good for the long term but we start from where we are)

Which in turn leads to management commitment to the Benchmark

Page 11: Paul Goodman Paul.Goodman@metagroup.com Introducing Software Metrics via Benchmarking.

11© 2004 META Group Consulting 01252 819494 metagroup.com

Devils and Detail – Short circuiting the “metrics debates”

Benchmarkers rely on comparability of data across organisations

Benchmarkers utilise standards wherever possible (IFPUG V4.2, Mark II, CMM-I etc)

Base metrics definitions are provided to the client as part of the deal – these are not debatable as comparability would be lost (although support for mapping client definitions to the Benchmark definitions should be provided)

Benchmarkers should be able to support (argue for) their definitions – i.e. they should be workable!

Page 12: Paul Goodman Paul.Goodman@metagroup.com Introducing Software Metrics via Benchmarking.

12© 2004 META Group Consulting 01252 819494 metagroup.com

Devils and Detail – Short circuiting the “metrics debates” - Downsides

Benchmarkers cannot be too innovative within their benchmarks E.g. there is currently little benchmark data

around based on COSMIC Some of the hiding places are suddenly not

there (you mean you don’t know how many defects were reported during warranty?)

This can lead to overt or covert resistance to the benchmark which needs to be managed

Page 13: Paul Goodman Paul.Goodman@metagroup.com Introducing Software Metrics via Benchmarking.

13© 2004 META Group Consulting 01252 819494 metagroup.com

Avoiding the Ivory Tower

The benchmark depends on a benchmarking model and two data sources:

The model: This may be defined within the benchmark

offering as non-negotiable – management will have been involved in accepting this during the sales negotiation

Definition of the model may be part of the benchmarking process (Meta ADM Benchmark) – involving client representatives

Page 14: Paul Goodman Paul.Goodman@metagroup.com Introducing Software Metrics via Benchmarking.

14© 2004 META Group Consulting 01252 819494 metagroup.com

Avoiding the Ivory Tower

Data: The benchmark database from the benchmark

supplier Project and Application Support data from the

client Typically the latter is less readily available than

the client may have believed This often leads to a recognition that basic

management information is lacking It usually needs the involvement of the client

Project Managers to get the data into the benchmark

Page 15: Paul Goodman Paul.Goodman@metagroup.com Introducing Software Metrics via Benchmarking.

15© 2004 META Group Consulting 01252 819494 metagroup.com

Avoiding the Ivory Tower

It is common practice to establish a Benchmarking Steering Group

This meets regularly under a defined Terms of Reference

Acts as a Measurement Co-ordination Group for the Benchmark

Will become a driving force for measurement within the client organisation

The Benchmarkers process should be well proven with other client reference sites

Page 16: Paul Goodman Paul.Goodman@metagroup.com Introducing Software Metrics via Benchmarking.

16© 2004 META Group Consulting 01252 819494 metagroup.com

Avoiding the Ivory Tower

The benchmarking process should incorporate the collection of contextual project information

Avoidance of ivory towers is achieved through: Involvement of the client in the process The use of a tried and tested process The Benchmarkers experience of applied

measurement in the real world

Page 17: Paul Goodman Paul.Goodman@metagroup.com Introducing Software Metrics via Benchmarking.

17© 2004 META Group Consulting 01252 819494 metagroup.com

From Data to Information

The primary focus of the Benchmarking initiative is on the resulting report to management

This report communicates not simply the raw data but meaningful management information, e.g. Comparative costs Productivity Time to Market Delivered Quality Process Maturity

Page 18: Paul Goodman Paul.Goodman@metagroup.com Introducing Software Metrics via Benchmarking.

18© 2004 META Group Consulting 01252 819494 metagroup.com

Managed Expectations

It is a fundamental rule that the Benchmarker should manage the expectations of the client with respect to: The scope and limitations of the benchmarking

model The use of a transparent process Involvement of the relevant stakeholders

Page 19: Paul Goodman Paul.Goodman@metagroup.com Introducing Software Metrics via Benchmarking.

19© 2004 META Group Consulting 01252 819494 metagroup.com

All is not perfect

During the first benchmarking cycle issues will arise These are likely to include:

Peer Group identification making the client think about what they do:– How is work categorised?– How is this work managed?

The client realising that some basic data is not as readily available as initially assumed

The growing recognition on the part of the client that much of the required data would add value if available on a more regular basis

The latter point can lead to a more general measurement programme requirement!

Page 20: Paul Goodman Paul.Goodman@metagroup.com Introducing Software Metrics via Benchmarking.

20© 2004 META Group Consulting 01252 819494 metagroup.com

Programme Set Up

The client can mitigate some of these issues by: Working to identify an agreed (by internal

stakeholders or the client and supplier for outsourced services) Application Portfolio

Establishing the set of “live” projects Ensuring that data (including plan data as well

as actuals) for recently closed projects is retained and is readily available

Establishing the concept of a warranty period for completed projects

Page 21: Paul Goodman Paul.Goodman@metagroup.com Introducing Software Metrics via Benchmarking.

21© 2004 META Group Consulting 01252 819494 metagroup.com

Moving Beyond Benchmarking

When the client and the Benchmarker work together to achieve a successful benchmark this often generates the recognition of the added value of metrics data

The Benchmarker will often provide advice on widening the benchmark initiative even if they are not directly involved

The Benchmark Steering Group can be a key catalyst in moving things forward

Client recognition that the measurement function needs resources

Page 22: Paul Goodman Paul.Goodman@metagroup.com Introducing Software Metrics via Benchmarking.

22© 2004 META Group Consulting 01252 819494 metagroup.com

Moving Beyond Benchmarking

That software metrics becomes a natural component of software development, support

and enhancement activities within the organisation

Benchmarking truly succeeds when it evolves to become the client’s measurement programme

The Goal