Something’s missing in my life

“Everything's going like I planned so far

But something's not there

That should be there

I gotta find it somewhere”

Isn’t it amazing how song lyrics can fit into some of the deepest concepts we need to consider? My memory of this is the Marcia Hines version of the Jabara and Asher classic. That probably dates me more than I’d like to admit!

But… something -is- missing. And it something we all know about but rarely discuss. 

Picture this… everything is running smoothly. Production is going like planned and the world seems a happy place. What’s especially satisfying is the resource-grade control-concentrator reconciliation. It’s perfect. All in agreement with less than 5% variation month-on-month. It’s a mine geologist’s dream! Why, I might even get a chance to do some mapping and actually look at some rocks for a change.

Ah….

Are you waiting for the sound of the next crisis? No? Well… what if I tell you something is missing?

I want to talk more about that monthly reconciliation. That plus or minus 5% difference between the resource, grade control and the concentrator… what exactly does it mean? Is 5% a good number? Is it as expected? Should it be lower? Should it even be -shudder- higher? And should I be worried about it? Surely not… 5% is great… isn’t it?

We are creeping up of that missing something. It’s something you might want to find and if you do you might just make a material difference to the profitability of your operation..

When ever I talk to people about reconciliation I hear surprisingly similar stories. It starts with “no one ever worries when we are making more metal than plan, but I get hammered when we are below forecast.” This is such a common phrase that I’ve even coined a phrase for it…. Dunham’s Law…. Ok, I’ll admit it’s not a law and I’m probably not the first to recognise it but allow me a little latitude. This one phrase is the opening salvo in a cascade of events that can lead down a very deep and dark hole.

Dunham’s law states: when no one is as concerned with over performance as they are with under performance, it indicates a fundamental misunderstanding of statistics.

Simple. 

My typical response to that opening statement is to innocently ask, “what makes an acceptable reconciliation result?” There are two typical answers, and they are both delusional…. 

The first is, “we are ok as long as we get a bit more than we predict - no one complains about that.” The second is, “if we are within +/-10% per month it’s ok - management doesn’t seem to worry about that.”

The amount of bad thinking wrapped up in those two answers never ceases to both amaze and infuriate. 

Here’s the implications of the first. That answer has a simple implication. If you don’t want your manager to complain alway under estimate. You’ve probably heard the mantra ‘under promise and over deliver’. That might make you look like a hero but is it the best thing for the business? Probably not. When you are persistently biased you are building losses into your system. That ‘under promise’ is a lost opportunity. Keep under promising and you might just end up locking in high costs or inappropriate capital expenditure in other parts of the business. An operation that is designed to deliver a certain result will be inefficient if reality is consistently biased when compared to the design. The downstream perturbations magnify - wrong fleet - wrong workforce - wrong concentrator - even wrong camp and flight schedule… you might feel like great because no one is chasing you and asking ‘where is the metal’ but as a strategy ‘under promise and over deliver’ carries high hidden costs. 

What about the second answer? The +/-10% band of acceptability? Surely that’s better than the ‘under promise’ alternative? Well… maybe… it at least plays to our cognitive biases. We humans like that 10% limit. It’s the most common response across a myriad of industries when managers and executives are asked about acceptable (and expected) error rates. 

Let me give you a hint. There’s nothing magic about 10%. It might be an expectation but it’s built on non-existent foundations. To know if 10% is a good tolerance limit we need to know a couple of things. Unless you want to be a sadist and slowly drive people to excessive drinking that is!

Firstly we need to quantify - with data - the capacity of the system. What is the natural variability of the forecast and production systems? It’s going to be different depending on the type of operation, the geology (imagine that!) and the people amongst other things. Why on earth would all those variable factors magically align on +/-10% for every business amd every commodity? They simply won’t. But measuring the error rate is hard work. So much easier to fall back on an unproven heuristic.

Secondly we need to understand -objectively- what level of variation is damaging to the business. That’s tied up with a complex set of factors including things like risk appetite, portfolio management philosophy, financing arrangements, and downright management fortitude and trust.

You would think this second point would be both obvious and easy… but it’s not. I mean, when was the last time you were explicitly told that your goal was to ensure cash flow within +/-10% (there’s that number again)? It is the sort of implicit objective that ‘everyone knows’ but no one understands. There’s a cost to achieving that goal and yet it’s not something I’ve ever seen explicitly stated! I do have one client who likes to ask if their plan is ‘P80’, they even have a process to assess the ‘P80’-ness of their forecasts…. But it’s not a valid process. It is only assessing if individual steps are, in the opinion of the expert, at an 80% likelihood. Their process ignores variability - especially variability between operational silos.

I once discussed this very point with a senior finance executive. My question followed a fairly brutal round of budgeting where 18 different ‘plans’ were iterated before one was accepted. I ask “what do you do with this budget cash flow now?” His answer was like tipping an ice bucket over your head… “oh,” he said, “we don’t believe any of it. We cut the forecast by 30% and use that to arrange our debt facility’.

30%. Well at least it wasn’t 10% again.

To be fair, I suspect that 30% figure was in fact, a good answer. After all, those 18 ‘plans’? Well they were not much more than shuffling numbers in a spreadsheet in a kind of complex goal seek - looking for an answer that satisfied the COO.

So we have two questions we don’t ask nor answer. What is the best variability I can achieve and what degree of variability hurts my business?

But it’s not these two aspects that are the real tragedy here. There is something else missing in my life…. 

What if I told you we could have 99.9% agreement between the resource model, the grade control estimate and even the concentrator and yet we could still be making a huge mistake? There really is something missing! In fact there’s more than one ‘something’, there are several but let me look at just one.

Here it is… the missing element. That missing aspect of reconciliation that is never reported or debated. It is never questioned. Have you guessed?

It’s those pesky waste dumps. 

We don’t measure the grade of waste dumps or even those long term low-grade dumps that are left for treatment when mining is over. We don’t measure an entire production stream and, without that measurement, there is no way we can approach reconciliation as a closed system. It is alway leaky.

We know this. If you’ve ever processed long term low grade dumps, or retreated ‘waste’ dumps left from a previous phase of mining, you know there are surprises. The grade of those dumps is all too often higher than anticipated. It may not be a consistent trend but there will be patches where the grade is surprisingly good.

Misallocation is unavoidable unless you are using a zero cut-off.

Where there is misallocation there is error and without measuring the size of that error we don’t know if our models are reliable. We could be ‘meeting budget’ and yet still be sending a large volume of economic material to the waste dump. Things would look great but the lost opportunity could be very very large…

Yes, something -is- missing. 

Not actively measuring the grade sent to the waste dump is akin to a concentrator no measuring it’s tails grade. So why do we berate the metallurgists if their tails samples are poor? At least they are trying! At many mines there is a strong focus on minimising tailing losses. Samples are taken, analysed and examined. Mineral liberation and grind size are checked and research is conducted into ways to recovery every last potential.

It makes you wonder why we don’t have a similar focus on losses to waste when we are in the cut and thrust of mining… the only strategies I’ve seen purporting to attempt to ‘minimise’ misallocation have relied on lowering the cut-off or, manual intervention by ‘ore spotters’…. I’ll discuss that questionable practice later!

I’d argue that the lack of focus is entirely related to a fundamental blind spot. We are so accustomed to -not- directly measuring the grade of -any- mine production stream that we don’t even see the risk…

Something missing? You bet! Isn’t it about time we changed?