There can be few things more satisfying in a work context than turning around a team delivering poor quality. It is such a virtuous circle: the customer gets a better outcome, the team feel better about the quality of the work and the reduction in rework, remediation, handling complaints etc. greatly reduces the cost burden.
When faced with a quality problem, the question is where do you start?
I’ve always found it useful to understand the cost of poor quality. In many cases, what appears to be a just a bit of a headache, soon escalates into a full-on call to action. The cost of poor quality typically gets broken down into four categories:
- Prevention Costs e.g. error proofing, validation, preventative maintenance, quality training
- Appraisal Costs e.g. checking, quality control
- Internal Failure Costs e.g. rework, backlogs, system downtime, latency
- External Failure Costs e.g. complaints, remediation costs, operational losses, revenue leakage
Many organisations don’t capture costs in this way and even if they do, they do not connect them to the same root cause. Quantifying the costs in this way soon casts the problem in a whole new light. A typical organisation spends most in categories 2 and 3. Recent announcements by a number of the financial services organisations in Australia attest to just how expensive external failure costs can be. But if you want to reduce the total cost of quality, the mindset needs to shift to a focus on prevention – and this is where Poke Yoke comes in.
For those familiar with Lean, the term Poke Yoke will need no explanation. For everyone else, it basically means mistake-proofing i.e. designing a process in a way that an operator cannot inadvertently make an error. The techniques are relatively common in everyday life, from speed bumps on a road that force cars to slow down, to lights in airplane toilets that only come on when the door is locked, to protect privacy.
It is an extraordinarily powerful – and in most cases low cost – way to improve quality. It can, however, be quite tricky to implement in a services environment, so here are a few thoughts. I’ve broken the examples down into four categories:
- Desktop Techniques
- Flow & Queue Management
- Quality Culture & Other
It’s a fairly long list still I’m sure I’ve missed out many things so please feel free to comment on your own preferred techniques.
For many staff in services their role is based around a computer. One of the issues here is that mistakes are hard to see – it’s not like a chocolate bar production line, where you know you have a problem when you see a pool of chocolate on the floor.
- Validation: one of the most obvious forms of fail-safe is to ensure fields on electronic forms are restricted, where possible, to allow valid data only e.g. an input mask for a telephone number, drop-down boxes and pick-lists.
- Dual Blind Keying: rather than have one person input data and a second check the input, have one person key and a second re-key critical fields only to check they match – a good way to maximise input quality. A variation on this theme is to have one person read out the details to be inputted, one person key and then swap roles to playback what has been keyed and compare to the original – time and resources intensive but depending on the consequences of getting it wrong – it may be by far the cheapest option.
- Green screen masking: limiting the fields that can be accessed is another reasonably common approach, including a lo-tech version using a sheet of paper taped over a screen with slots cut out in the area where data needs to be keyed.
- Spell and Grammar Check: it seems obvious given how often we use these tools, but the trick is to switch them on as a default!. Make sure the language is the same across all centres and teams performing the process using, for example, the same regional form of English – US, UK, Australian etc and that all team members have auto-check set as the default.
- Alarms: this can be anything from setting reminders and notifications to pop-up messages for critical tasks. An example might be: as a final check before sending an email, which could potentially create privacy breaches, a pop-up box might ask: “are you sure you want to send x data to person y?”
- Dual Monitors: these are relatively common now – one monitor displays the work request and the other is used to process the request. This avoids having to remember what is being asked. It can also help where team members require access to multiple applications. Setting out the “windows” in a logical flow reduces the risk of pasting data from the clipboard to the wrong application.
- Generic Mailboxes: a common request from premium clients is to have a named contact. The issue here is that emails may go unanswered if the relevant contact is ill or on leave. Generic mailboxes are far safer and using the prioritisation tips above will ensure the right team member is allocated to the task.
- Structured Input: where possible, capture data in forms not unstructured emails and preferably with typed data not hand-written. Not only does it reduce the risk of errors, it also dramatically increases OCR read rates. With inputs that include images, ensure the resolution is set correctly to allow operators to zoom in without the image becoming too blurry.
- Templates: an obvious solution for teams that are responsible for communicating with customers – use stock templates (where applicable) that can be auto-populated or have validated fields to personalise. This is a great way to ensure consistency of language, tone, grammar etc. It is particularly useful where team members are not native speakers of the template’s language.
One of the most common causes of errors is a lack of standardisation. Driving standardisation is not easy but it always starts with good documentation.
- Process & Value Stream Mapping: documenting processes and overlaying the failure points and relevant statistics is usually as good a place to start as any. Visualising a process nearly always reveals the blatantly obvious that somehow isn’t quite so obvious when you’re in the process day after day
- Work Instructions: as ridiculous as they may seem, one of the simplest fail-safeing techniques is to actually write down what you expect a team member to do (in detail). The mere process of getting all of this information out of the experienced team members’ heads will not only help get new hires up to competency far faster, making far fewer mistakes but also provide significant benefits in simplifying and standardising the process. Without a reason to talk about what they do and a “system” to enforce process adherence, each team member will naturally create their own version of what to do.
- One Point Lessons (OPLs): OPLs are ideal for tools such as video conferencing, phones, photocopier, printers. Ideally they are a simple infographic on a piece of laminated paper saved next to the device which provides relevant work instructions in how to use the device.
- Tags: using tags to make searching for documents easier is extraordinarily simple fix to the perennial problem of thousands of lost hours trawling through shared folders. Ensure you have a standard library of tags and train your team to use them, not make up their own.
- Checklists: I have written on the many benefits of checklists before and cannot overstate how critical they are. Most importantly make sure your team members performing the same process share the same checklists and when one team members finds a way to improve it all benefit.
- Training: training is an interesting one. On the one hand, in an ideal world, the process should be so simple and so intuitive that people (customers or team members) don’t need to be trained to operate the process defect-free. That’s obviously behind the intent of techniques like human-centred design. However, for most organisations with complex legacy processes, training is not only essential, it is a moral obligation to your staff. Anyone who has stared helplessly at a screen willing the answer to manifest itself because they are too embarrassed to ask their supervisor for the umpteenth time what to do will recognise just how crucial training is. Remember, team members move on to new roles so training is not a set-and-forget activity rather a continuous process of skilling and embedding knowledge.
Flow and Queue Management
Many of these techniques are oriented around ensuring you meet your service level objectives. Given the fluctuating nature of demand, this is one of the trickiest problems to solve – the Goldilocks service – delivering on-time – not too late, nor too early, but precisely when the customer expects you to deliver the service.
- Prioritisation: most teams have prioritisation rules for specific clients, segments, work types, due date etc and many workflow tools allow you to make these settings in the system so that the highest priority work is always at the top of the queue. Even the rules capability in most email applications allows you to set up pretty complex prioritisation rules. But if you don’t have a system-enforced way to pop the most important task to the top of the queue, simple things like using coloured paper or different coloured inboxes will help reduce sequencing errors.
- Colour Coding: in both paper and electronic environments setting the background colour of the “paper” for different customers can help with prioritisation.
- WIP Windows: essentially a simplified combination of Little’s Law and some statistical process control can quickly help you draw charts that set upper and lower level control limits for what is an acceptable level of WIP (queue depth). As long as your queue is within the control limits, you can expect to meet your service levels (under normal operating conditions).
- Kanban, Short Interval Control & Heijunka: these are a range of similar techniques to ensure you smooth out production and get through your work by breaking the load (number of work requests) down into smaller time intervals – instead of thinking how many do we need to do today, think how many do we need to do between 9-10 am. Checking in at the end of each shorter time period allows you plenty of time to take corrective action if you get behind. Heijunka boxes (essentially stacked in-trays), with each inbox representing an hour’s worth of work are a great way to manage throughput and pace.
- Visual Controls: there is a mountain of work written on this topic – essentially, make the work visible and the team will respond when the problems become obvious – this means dashboards, wall charts, tickertape displays and a whole host of other techniques – just make sure everyone can see them and they’re not buried in the Manager’s computer. Creating physical representations of the workflow can have a dramatic effect – I’ve seen clothes racks with “pegged” work requests hanging from them. All of the information was in the system, but, with the clothes racks in the middle of the floor, everybody could track progress easily.
- Runners, Repeaters, Strangers: whilst typically associated with optimising throughput and flow, this technique of triaging work requests and breaking them into simple requests, more complicated requests with known variants and the really unique, highly complex requests, means that you can allocate resources based on skill level and don’t put less experienced staff in the unenviable position of working on requests beyond their current capability.
- One Front Door: many quality issues arise from upstream teams choosing their own entry point into a process. This may be sending a specific work request to someone they know in the general function not to the capture point for the start of the process. All work requests need to enter at the start of the process not half-way through.
Quality Culture & Other
At the heart of a quality culture is the simple fact that people care about the customer and take pride in their work. They are empowered and equipped to solve quality problems as they arise and do not walk-away from a problem. Disneyland is a great example where keeping the park clean is everyone’s problem If there’s a piece of litter on the floor, the nearest member of staff wil bend down to pick it up whether they are the most junior or senior member of staff in the park.
- It’s Everyone’s Job: the most critical element here is for everyone to recognise their role in delivering a quality service – it may not be my fault, but it is my problem. It’s doesn’t matter which function you are in, you have a role to play in making the service defect-free. Typical examples include sales staff sending in customer orders that are incomplete or inaccurate, technology providing services with unacceptable levels of latency or drop-outs, product teams teams designing products with manual work-arounds.
- Don’t Pass On Poor Quality: if you receive a work request that has defects, under no circumstances do you keep working on the request and perpetuate the error down the line. Fix the defect by returning the request to the source or contacting the supplier of the defect to fix the issue and then solve at source. There are obviously practical and policy challenges with potentially contacting customers, but the principle applies and this needs to be addressed.
- Provide Space: one of the major causes of human error in services is associated with the operating environment. Expecting staff to multi-task, accept interruptions, rushing them will only lead to defects. Create a safe environment that is conducive to delivering quality output.
- Lean 5S: a place for everything and everything in its place. The lean toolkit provides a range of tools to ensure the physical environment is set up for success; clean, uncluttered, right tools to hand etc.
- Preventative Maintenance: whilst this may sound like a manufacturing or technology task there are many ways that it can be implemented to ensure the Ops environment is set up to minimise failure. For example setting regular cycles for restocking the printer with paper and toner, making sure there are always adequate stationary supplies, swapping out computers and laptops at the end of their useful life before they break down, having chargers available in meeting rooms, routine maintenance for tele- and videoconferencing equipment etc.
- Redundancy: most people will be familiar with the idea of redundant capacity from a technology perspective eg two comms cables coming in and out of a building in case one gets cut, in an ops environment this can mean having spare laptops, and peripherals or using SWAT teams to provide spare capacity if a specific team is hit with sudden turnover or illness.
- Multi-site: running a multi-site operation is an extension of the redundancy concept. If an operation has 3 locations running the same process and loses one eg due to adverse weather, transport strike etc, the other two should be able to pick up the slack – at least for a short time. It does make it more complex to manage however.
A long list, but hopefully useful. The main point here is that prevention is better than cure. The people working in the process are the real knowledge workers, they have the imagination and the process understanding of what to fix and how to fix it. Management’s role is to empower them to do so, create a safe environment for them to succeed and then get out of the way.