Service level agreements get broken all the time. That’s as true in the physical economy as it is in the digital one.

I learnt a lot about service levels when I worked in a coalmine.

At one level, coal mining is simple. Coal is valuable. The shale that surrounds it isn’t. Mining is about digging out as much coal as possible while leaving the shale behind.

In practice, things are a little more complex  It’s often hard to dig out the coal without shifting a lot of shale, and the boundary between the two is often unclear – strata that are 90% coal and 10% shale grade gradually into strata that are 90% shale and 10% coal. 

And predicting the exact shape of the coal seam as it meanders through the earth requires a lot of skill. But what makes things really complex are the contracts, the mine’s equivalent of service level agreements.

The mine I worked in had contracts to provide coal to the local power station. These had tight specifications about the quality of the coal we must provide.  For example, they contained substantial penalties if the ash content (the amount of shale mixed in with the coal) exceeded a certain limit.

So this meant we were very careful not to mine coal that exceeded the specified ash content, didn’t it?

Not at all. The mining engineers spent most of their time working out how to mine as much shale as possible.

One reason shale isn’t valuable is that there’s lots of it. Far more than coal. So the mining engineers wanted to sell as much shale as possible to the power station. Doing this increased the mine’s coal reserves dramatically.

They could do this because the penalty clause was only invoked at a certain ash level. Say that level was equivalent to pure coal with 15% shale mixed in. Then selling pure coal made little sense, you had to throw away all the slightly impure coal, yet received no bonus for exceeding the required quality level. 

So we made sure there was always enough shale mixed in with the coal to just meet the required level. This meant we could mine more rock and call it coal.

More than that, we often tried to mix in even more shale (or at least impure coal) than the quality level allowed. Sure, we had to pay penalties, but we could also sell a lot more rock. 

Selling twice as much rock at three-quarters of the price, for example, can generate a lot more revenue and profit than selling a smaller volume at full price, provided your costs are properly controlled.

This is where the mining engineers came in. Their job wasn’t to maximise the quality of the coal we sold. That would have been relatively easy. Their job was to calculate the optimum trade-off between the volume of “coal” that we could sell, the price we could get for that coal, and the cost of mining it. 

In most cases, this meant breaking contractual quality agreements and paying penalties.

But there’s no coal in the average datacentre. How is this story relevant to digital services?

The service level agreements we negotiate with our various partners are just like that coal mine’s contracted quality levels. They’re a starting point from which the service provider will design its systems and processes. 

They’ll try to meet them if they can, but not at the expense of levels of staffing or redundancy or complexity that cost more than the cost of any penalties for breaking the SLAs. That’s not because they’re bad people, but because this dynamic is built into any contractual negotiation.

So how do we deal with this when we’re buying digital services?

One route is to build more “sophisticated” SLAs. We can create incentives for exceeding the required service levels. We can invoke penalties on top of penalties. But this is an arms race – people will always find ways to game the system. 

We end up simply creating complexity, which in turn creates additional costs (not least for the lawyers).

The better route is to keep our service level agreements simple, and to be on the lookout for the unintended consequences they may bring. 

Focus your energy on building a strong relationship with the service provider (relationships trump economics in most emergencies), and on building your own systems and operations to accommodate the occasional service level failure.

That’s what the power station did. Its engineers had been around long enough to know that the coal quality specifications would be broken. They built their furnaces to accommodate more ash. 

They timed their service schedules to clear out the ash before it built up too much. They set the penalty fees at levels that would pay for these activities.

Remember this: service level agreements provide a starting point for operational planning, not a rigid line that will never be crossed.