Tuesday, December 25, 2018

On tenders

There's no end of advice on calling and evaluating tenders for building/construction services, whether professional or trades, it would seem. This is suggestive of a problem not yet solved.

One of the most popular approaches is to seek tenders that satisfy listed performance or capability criteria and achieve the lowest price while demonstrating high performance. Quite a tension!

In the UK the desire is to fine the Most Economically Advantageous Tender ("MEAT") where price and performance criteria are melded in the hope of representing a meaningful measure of overall benefit to the principal.

As in other places, the performance criteria are usually given some type of qualitative or impressionistic but hardly objective, repeatable or reliably reproducible score. The various criteria are then 'weighted' by an equally subjective and probably unreliable method. The 'dartboard approach' to tender evaluation.

The prices are 'normalised', removing information, and the result is multiplied by the weighted performance score and it is imagined that knowledge is produced!

Patrice Fabien has written on this topic with some insight into its limitations.

I'd like to propose a better way than the hit or miss that is common.

The performance criteria need to be characterised by sub-criteria so that the evaluation can be objective, with scores given on the basis of countable items. For example, running from:

0 = meets no sub-criterion up to
5 = meets all major and minor sub-criteria.

The weighting is best achieved using a system such as pair-wise analysis, as explained in the MITRE STEP methodology, and precisely in the evaluation system.

In my view, this produces a reliable and reasonably objective performance or capability score for the project performance.

This score is then subjected to a threshold for progression to the price evaluation.

The threshold might be set as an absolute minimum acceptable score, or minima that must be achieved on a number of criteria. Thus a low score on identified criteria will eliminate the tender from further consideration. After all, there is no point considering an offer that fails to deliver, no matter what the cost!

The next step is the price comparison.

Only those scores that meet or exceed the threshold should be considered at this point. If there is a large field over the threshold, a 'countback' through the scores might be used on a criteria basis (tenders with lower scores on critical criteria eliminated or only the highest 'n' scores considered.

Rather than mathematically normalise the prices (which destroys important information) to produce a 'score' then multiply this and the performance score, it is preferable to proportion scores against the lowest 'above the threshold' price.

This proportions the price by performance, in effect, pricing the performance.

The result looks like this:

Sunday, December 2, 2018

Good Design

Lots of talk about 'good' design. But precious little on its parameters. Sometimes the 'good' transmogrifies from effectiveness to moral rectitude, touching other way-stations on its travels.

There are a few lists around of dimensions of good design.

One of the most famous is Dieter Rams' 10 principles by which good design is:

  1. Innovative
  2. Makes a product useful
  3. Aesthetic
  4. Makes a product understandable
  5. Unobtrusive
  6. Honest
  7. Long-lasting
  8. Thorough down to the last detail
  9. Environmentally friendly
  10. As little design as possible
Some of these are meaningful. Others are empty, arbitrary or judgemental. For instance, "honest." Really? "Unobtrusive"? Not good if you are designing the entry to the Emergency Department of a hospital. "Aesthetic"? Aesthetic what? Good, bad or ugly...they are all 'aesthetics'. Aesthetics is about beauty, it doesn't mean beauty itself. "Environmentally 'friendly' ". Peurile and unhelpful.

Archdaily sets out a little more description on this same list.

Here's another attempt.

Parsimoniously fit for purpose.

Easy to say, hard to do. That covers 1, 2, 4, 7 and 9 of Dieter's list.

Let's look at the coverage:

1. "Innovative" without purpose is waste. The wheel is good, but we innovated to tracks when they were better. Tracks on a racing car would be no good! Innovation for its own sake without a driver external to the object of attention is also waste; and possibly frivolous. On the other hand, if it takes us to unexpected benefits or stimulates external unexpected benefits, opportunities and investment, then all the better.

2. "Makes a product useful"

Of course, 'fit for purpose'. Noting that the wider 'purpose' is cast, the more better (!) the design is. The best design is produced on the back of deep understanding of all the drivers of purpose. These can range from an investors' objectives, to users' objectives, usability, lifetime costs, usable life, the applicable objectives of others who have in interest in the subject...and so it goes.

4. "Makes a product understandable"

Same as previous.

7. "Long lasting"

Better: lasts as long as necessary. A stage built for a temporary festival that would last 1000 years is not what we want...although some stages build for permanent use have lasted well over 1000 years.
If it lasts too long, the risk of over use of resources or over investment in the subject occurs.

9. "Environmentally friendly"

I've never known what this means, nor does anyone else, I suggest, as it is used to promote all sorts of political and commercial agendas.

Here's what we are after: Meets environmental requirements and in an overall sense, parsimiony of materials, effort, operation and ownership leading to minimal consumption of resources: just enough consumption to meet the user need, and just enough consumption through maintenance and cleaning. There is also a cost-performance trade off with any measures that are not core to the purpose of the design subject. Over-investment in one 'environmental' outcome will reduce the capacity to achieve other environmental or general product outcomes.

Uses inputs (materials, components and execution techniques) compatible with requirements.

This meets Dieter's 7, 8, 9, 10.

The core parameter is cost-effective deployment of industry capability. This can include innovations that are brought about, of course, but it is hardly good design if a product is impossible to achieve economically or with other consequential 'dis-benefits'.

Nothing meets number 6. To help think about 'honesty' in design, compare the 'honesty' of a CT scanner without the cover. Honest? Helpful for an anxious patient? Easy to keep clean? I don't think so, and darn scary when it starts spinning.

Enjoyable to own, use, operate.

Dieter's numbers 2, 3, 4, 5, 8.

Enjoyable is hardly the word; perhaps enjoyable for the use: a hospital foyer's 'enjoyable' would be able to reassure patients, encourage staff and attract visitors. An opera house foyer: enthrall, excite, stimulate?

Enjoyable relates to practicality, affect, and aesthetic perception appropriate in the product's context.

Thus, in keeping with Dieter's number 10 my list is 3, his is 10. Less is more.

Another take on lists is my five item version:

In every way suiting the needs of the users, owners and investors: even the needs they don't realise for functional fit, comfort, future adaptability, etc.

Again, in every way: value for money in selection of materials, systems, components, operations, maintenance and future adaptability. This is an umbrella 'economy' including use of energy and maximisation of passive climatic adaption.

No mucking around by the builder and his supply chain to realise the design.

As above in the list of three. This can go from 'merely' comfortable, to spectacular, depending on the building's role.

This is as broad as you need it. As one uses, passes by and/or views the building, what it does, is and provides to the place, the user and its viewers is comprehensible. That is, the building represents its performance 'strategy' to the user and community. This can extend to contextually adapted (a 'good neigbour'), 'coded' for its purpose by the visual/tectonic and spatial language of the design, tells the user what the user needs to understand: where one does what. For example the entrance presents itself to the user for comfortable and helpful usage (back to 'usable'). An opera house, conveys that it is a special, fun and enjoyable place to be, etc.

Saturday, August 18, 2018

Programs and programs

My recent work has included a lot of interaction with human services businesses. They use the word 'program' differently from its use in generic project management; indeed, even the public sector based Prince2 system conceives as programs exclusively in such terms too. (in the MSP system).

In human services, 'program' when it is not applied to infrastructure, for example, the hospital construction program, means, not a collection of related projects, but a systemic intervention to bring change in people's lives; the 'people' are usually known as the 'target group' (in the Army we used 'target' to mean something we would obliterate...not so here, of course!).

These programs are based on what it terms a 'program logic model'. This sets out how the program will work to effect the results in peoples lives that it sets out to achieve. Conceptually much like an infrastructure project.

The PLM works back from the results to 'Inputs' via (in backwards order): Outcomes, resulting from Outputs, resulting from Activities, detemined, or defined by Resources or Inputs selected on the basis of Need or Opportunity.

Again, there is an obvious resemblance to the typical project course.

Where the major difference is is in the Activities-Outputs-Outcome set. This trio is a constant cycle that operates for the duration of the program, and may be a permanent operation, for all practical purposes. For example, the 'road safety program'.

The PLM is not only a plan (at high level) but a means of inquiry into the program; so it forms the basis for definition, specification, design and delivery; much like a schedule; but brings together schedule, WBS and delivery in one unit.

A typical PLM is below.

Wednesday, August 15, 2018

Top level project report

I posted this comment on a recent Linkedin PM forum, to the request for the items that a CEO report should contain for a project.

At the very top level: the current forecast outturn cost, compared to current budget; the current forecast completion date, compared to current approved completion date; earned value status, dollars committed, number of unresolved issues in an aged table (e.g. unresolved for 2 weeks, 1 month, 2 months) and current value at risk.

 Now for some detail.

The critical information for a CEO is how much will this cost and when will it be finished. This 
would be compared to either the business case or the current approved state of the project.

Reports should have a graph of all previous forecasts to show trends.

 A subsidiary interest would be the expected NPV of the project compared to the business case, or current approved NPV. For some CEOs this would be critical. Indeed, for all, it should be!

Earned value status is an obvious one; but if a formal EVM system is not used, it can be given an approximation: how much was estimated to be spent for a certain state of production, compared to how much was actually spent. If the contract has an estimated cash flow, this could be an input to tracking this.

However, all these are backwards looking. The CEO would need to have information about the project's future.

Two proxy indicators that could be reported are:
  • dollars committed (that is, subject to contracts with suppliers and sub-contractors). Comparing these against cash flow projection can be used to assess the project productivity.
  • aged tracking of issues from the issues log. The number of  open issues over time would give an indication of the general project state and performance. If notified issues won't do it, then it could be decisions pending, or contractor claims unresolved (i.e. yet to be agreed, accepted or rejected).
The final item is current value at risk. I'm assuming here that the project has a quanitified and costed assessed risk. As the project advances, some risks retire, and the final project cost becomes more secure in a tighter range, but others may emerge, change or be costed more accurately. This might lead to an increase in either the final out-turn cost or the expected range of that cost (e.g. project is 80% certain to cost between $10m and $11m). 

An example of a similar report format is below, for illustration. It is also worth referring to the One Page Project Manager for another, more operational, approach. 

Tuesday, June 12, 2018

Confuse me with lifts

I'm sure you've seen it; maybe even done it.

Hit the wrong lift door button when in a hurry. Maybe closing the doors on someone hurrying to hop on.

It's not hard. The Australian Standard symbology is confusing as it depicts door states in conflict: one shows the desired state of the door being closed: the arrow heads pointing to the door leaves closed, suggested by a vertical line; but the other shows not the desired state, but the current state: another vertical arrow, but not the state we want for the doors open. At a glance they are impossible to distinguish promptly.

A better symbology is in the lifts at a Council Chambers I recently visited in Sydney.

Tuesday, May 8, 2018


I've long had a bee in my bonnet about stair design. After centuries of making stairs, we still seem to not be able to get it right reliably.

When I find a comfortable stair, I measure the rise and tread.

At a weekender I've used, the stairs were a comfortable 170 rise and 285 treat.

Here's the stair at Wahroonga railway station in Sydney.

Rise 145mm, Tread 360. Very easy to use.

Saturday, November 11, 2017

Project software

A run down on the project software I've used over my career so far.

First in NSW Public Works Department

Vision, running on an Alpha Micro minicomputer (size of a large dishwasher)
Vision ran in two parts. First load was of activities ('tasks' is the word amateurs use), and the activity application was run. Then the activity relations application was run to weave the activities together.

It printed out at great length as you could imagine for a time scaled network diagram.

Next package was Timeline 3 running on DOS.
I used this in a medium sized architectural office where I was senior architect. I was running a small job to design a new factory for BHP, making welded plate beams. The firm seemed to have little idea of job pricing, so I created a schedule in Timeline, put costs against all the staff and consultants we'd be using. I printed this out on my little dot-matrix home printer and showed the partner in charge. He was struck by the cost of the project. I think it was the first time he'd seen such a thing.

After this I upgraded to Timeline 4 on DOS as it had macros.

My next package was CA Superproject in Windows.
This was a charm to use, although I was irritated by it showing each resource as a separate activity bar. Took up lots of space on the screen, and lots of paper to print. It may still be available at Vetusware. I might have some disks somewhere as well, but would be surprised if they still worked.

I didn't touch PM packages for sometime then, until I was working for a multi-national design-construct business. We used Fast Track and an architectural information package, name of which escapes me (maybe it was Fast Track!). I saw an early version of Prolog (now owned by Trimble) at a trade show and was impressed that here was software that supported full management of a project, not just schedule and cost. It managed the entire information flow, which is the heart of PM, in my view. Details 3 is a similar package, but I'm not sure if it is still live.

Then I came to Microsoft Project, in a very early incarnation. Was good, and I liked it, although the presentation control (for printing) could have done with some tweaking.

I moved to another multi-national D-C firm where Primavera P3 was used: a great PM package. We also looked at Expedition for full project information control, but as we were implementing BAAN ERM, our efforts went there instead. This also distracted us from implementing Cobra for earned value management. We also had a stripped down version of P3 called Suretrak. It was a superb little package. I've got the disks for that somewhere too.

Primavera was great. It really felt and performed as a top of the line professional package (its now P6 and owned by Oracle).

I missed it when I joined a public sector operation and back to MS Project; OK but not great, although it has improved over the years.

One aberration I encountered in this public service organisation was during the implementation of SAP financials and HR systems. SAP projects was bolted on. This was so bad it wasn't even wrong! Project management as conceived by an accountant is not the way to go. It was so clunky and non-project in its concept-of-workflow that it was impossible to use for project management. Reporting, maybe, cost accounting, maybe, but not for PM. No one used it. The idea of tracking people and costs on the one platform was a good one, but force feeding by corporate types who know nothing about project management was a recipe for disaster, they not knowing what they didn't know, and thinking that multi-degreed engineers, architects and quantity surveyors, whose first degrees were twice the length of theirs had nothing to offer was risible.

Now, a few curios:

TurboProject. From the same stable as TurboCad, and pitched at the same market: I'd call it sub-professional. OK but not brilliant. The product seems fast and reliable with sufficient functionality to satisfy more than basic needs. I used it at version 1, and its still around, in a number of variants and would probably be great for a firm that did smaller contained projects: event management, publishing, simple buildings or civil works, etc.

Key Project Manager. A small 'task' management package; came on a magazine cover disk decades ago and I used it in Windows' early days. Simple for non-project tasks and even had task durations with calendar views to help coordinate. But here, as most 'task' managers, the fatal flaw is no way of assigning or tracking dependencies.

In the world of Mac the only offerings I am aware of are Merlin Project and ConceptDraw Project. Both are at the same level as TurboProject, in my estimation, but I'd be happy to be proved wrong. That is, good, but not for major projects with hundreds of activities, dozens of calendars, multiple baselines, earned value reporting, etc.

Update: a quick look at the Mac websites was good; both packages have come along since I encountered  them about 10 years ago. Quite mature products. Merlin has the ability to link lots of files to activities (they call them tasks, like a project is equivalent to cleaning the house). Now this is starting to be where we need project management software. The big job in project management is managing and communicating information. Tracking this torrent of what used to be 'paperwork' is at the heart of it. Tracking project configuration changes is also essential: drawing and specification revisions, scope changes, work remediation...the time schedule approach to PM is nowhere near this.