Monday 17 January 2011

How do you escape the “domino effect”?

In the second episode of Michel Roux’s Service, the owner of an Indian restaurant in Birmingham describes the traffic jam in the kitchen as a “domino effect”. Diners have been seated late, and have placed their orders late, so a bunch of orders come into the kitchen at the same time. They’re readied as quickly as possible for the servers to take to the tables. The servers struggle to get them all out as quickly as possible. But the net result is that a lot of people get their food later than intended, and the problems carry on through the rest of the evening.

The domino effect is familiar enough to all of us, but how does it work? It is in fact a potential problem in all systems. On the programme, the problem seemed to occur at “the pass” – the point where orders were coordinated between front and back of house. In fact, the problem started when a party was mis-seated. When the people who should have had their table turned up, this had a knock-on effect on another groups – a total of three groups were re-seated. This in turn affected the timing of the orders, which all came into the kitchen at once. This meant the orders were ready to serve too close together for the servers to get them out promptly. As the service backed up, more and more orders were being delivered late, affecting far more parties than those originally involved. This expanding ripple of service problems is what we call the “domino effect”, because one problem has a knock-on effect on every process step “downstream” of where the problem occurred. Often, the most visible effects of the problem are at this downstream end, and it’s not necessarily clear to those affected just how far back in the processes the dominos had started to fall.

I saw a beautiful demonstration of it at a coffee shop recently, and it will show you how this famous effect actually takes place, and what you can do to change it.

In the coffee shop in question, a branch of a major chain, I observed people being served coffee between 10-11am. They queued at a counter, where a member of staff took their order, gave them any food part of their order, and took their payment. They then moved on to the end of the counter, when they collected their coffee from the barista. The barista got her cue about what to make from the staff member at the till as the order was taken.

The barista could make a coffee in an average of just over a minute. In the course of 40 minutes, 29 orders were taken, so she had enough “capacity” in terms of time to make all those orders, with time left over.

Had the customers come in like clockwork every minute or so, and had all the coffees taken an equal time to prepare, she’d have probably done fine. However, clients don’t come in like clockwork – they come in bunches, with pauses in between. And some coffees are complicated, or get spilled. Also, there were small disruptions – from customers wanting to ask a question, from a member of staff who came through to do some cleaning, and from the staff member on the till having a complicated food order to attend to, so that she had to rush to catch up taking orders, and give a bunch of orders to the barista at once.

This is why the domino effect comes into being. When things work like clockwork, with very little variation, you don’t see it much. Where you see it is in naturally chaotic systems – for instance, those that rely on human behaviour and on chance. The natural variation in these systems puts pressure on the links in the service chain, and if capacity is tight for even a short period, it can have a knock-on effect (a domino effect) on the chain long after the initial cause is gone.

Here’s how my observations went. In the first 15 minutes, only three people came in, and all were served within three minutes of walking into the shop. Then, at 10:16, two people came in – separately, but within a minute. The first was served within a minute, and the second within two minutes, so they went away happy. Then at 10:18, just as the second was getting her coffee, three more people arrived. Unfortunately the first person’s coffee was complicated, taking a full two minutes. The second was normal, but the third one took 2 minutes as well, so this last person wasn’t served until six minutes after she walked into the shop.

Meanwhile, another three people had come in. Although their coffees were quick to prepare, the barista was already behind, so they too had to wait six minutes for their coffee. While their coffee was being made, another three people came in. And while their coffee was being made, three more people arrived. These people all had to wait 3-4 minutes for their coffee. The service times were also lengthened very slightly by the cleaning staff passing through, and a customer asking a question.

At this point things got really busy. Between 10.31 and 10:40, 13 orders were taken. This was more than the barista could deal with at the speeds she’d shown up to now. Three came in at 10:31 alone – and of course the barista had started out a few minutes behind. The first was served in 4 minutes, but all the rest took 6-7 minutes from walking in to getting their coffee.

As long as people don’t mind the wait, this would have been fine - but now the domino effect began. Some of these customers had small children who quickly got bored and started acting up. Some people were so keen to sit down that they went and put their stuff on chairs, before rejoining the queue, to reserve a place – which annoyed customers ahead of them who weren’t doing it. The seating process was becoming disrupted, and that was making clients unhappy.

The biggest problem with the domino effect is that it's most likely to cause service failures at the times when you have the most customers, so even if it happens rarely, it affects a disproportionate number of your clients.  Moreover, it affects people who had no idea what originally happened - to them, it just looks as if bad service is a normal part of your business.

If there is a domino, does this mean the service system is bad? No – but it does show that the domino problems have not been planned for. Every system with more than one process step will get domino effects when demand is high and there is a disruption to the regular working of the system. A good service team understands this, forsees what the domino effects may be, and plans what to do to stop the domino effect in its tracks.

Preventing the problem is not just about throwing capacity at it – that just eats up all your profits. Instead, the plan for preventing domino effects will be to focus on the spots where capacity is a problem, and change them to make it easy to stop the domino effect before it damages your business.

One approach is to tackle the bottleneck resource that has caused the problem. In the case of the coffee shop, the barista was the key to the speed of service. Once the coffee-making got really behind, the whole system was stuck until customers stopped coming. And no one wants customers to stop coming. So what could have happened was to expand the coffee-making capacity:

  • Batch processing: The espresso machine could make three coffees at once, and in some coffee shops, you will observe the barista taking advantage of this, setting up and running three espressos at the same time if there is a queue. While they are dribbling out, she can work on the other steps of the coffee preparation.
  • Multi-tasking: In other coffee shops, you will see more than one person working the espresso machine. Typically one person will be full time on it, and someone else will be full on the till, but there will be floating capacity – someone who can pitch in to whatever part of the service chain has got backed up, and relieve the pressure, while doing other things (like cleaning) when the pressure is off. Where there are several different services on offer – filter coffee as well as espresso, tea, hot chocolate and so on – this may be particularly effective.
  • Personal service: Still other shops will have one individual dealing with each customer – taking their order, getting their coffee, and so on. That way, if one customer is taking a very long time, the queue will keep moving because other servers will be dealing with the next customers. This is typically not an efficient way of working, because people will get more done if they are just doing one part of the process, rather than the whole thing. However, it may well be appropriate where there is very little in common between one order and the next, and where self-service is an option, for example in fashion stores.
  • Productivity: Finally, rather than dealing with the systems, many coffee shops would address the underlying problem first: the slow speed of coffee production. The barista in this shop was slow compared to some of those I’ve seen in busy urban cafes. Experience and an effort to make her motions speedier and more accurate will no doubt help this barista serve customers more quickly over time. 
Another approach is to deal with the effects rather than the cause. Here, the principle problem for customers was the combined wait in the queue and risk of not getting seated. Some shops deal with this by avoiding counter service altogether and seating customers before they are served. But even with a counter service, some coffee shops will start taking orders right through the queue, and telling customers to have a seat, and that they will be called when their order is ready.

Each approach will be better suited to some contexts than to others. The point is that the service team needs to have a well thought out plan that suits their business and their customers. They need to know in advance what to do, so that when a disruption occurs, they can take action straight away.

How could the young people in Michel Roux’s care have dealt with their own domino effect? The problem was in serving, so as with coffee shops, they had some obvious options:
  1. Used the order-taking process to even out the flow of orders going to the kitchen
  2. Move people temporarily from other roles to serving food to the domino tables
  3. Batched the serving, for instance by getting a trolley that would hold the orders for many more people than a server’s arms can deal with
Each of these options has advantages and risks. They didn’t spot the problem early enough for option 1 to be useful, and option 3 may also not have been possible given the equipment the restaurant supplied. In the event, the young people chose option 2 – they took their receptionist off the door once the restaurant was full. However, their biggest problem was that they appeared not to have a plan. They did not know what the knock-on effects of their actions would be, and the whole of the evening was spent in crisis management. With a plan, they may have made better decisions, and could also have told customers what to expect.

The domino effect is unusually damaging because it takes a small problem, such as a complicated cup of coffee, and turns it into a service failure for many more people who should never have been involved. Because its effects last far beyond the original incident, it is often hard for staff to remember what the original cause of the problem was, and the rush of dealing with its effects makes it hard to stop and think the issue through. However, by planning in advance with a system that will lessen the effects of a domino incident, businesses can help improve their service to all customers, and deal better with the variability that is natural to customer-facing businesses.

No comments:

Post a Comment