Sometimes when we start investigating issues related to rework and scrap on the shop floor we discover the real root cause originates in the office before production gets the order. A forward-thinking printing company wanted to kill-off one of their biggest profit drains – reprinting jobs. This company discovered that at least 50% of reprints (reworking is usually impossible once you have ink laid-down) were traceable to issues in the front office. But their problem was this: exactly where in the front office are the mistakes originating? Everyone had their hypothesis and the finger pointing from days past was recognized as an unacceptable approach. In addition, the company has the objective of reducing lead times – recognizing that too much time was pre-production – led to part of the improvement team’s charter. In this first of a two-part series of articles I will describe how the process improvement team went about documenting the current state of their process – and driving future improvements to speed and quality.
The company elected to form a cross-functional team and committed to finding all the sources of problems from the time the phone rings, to taking a new order up, to the time the art department got the final inputs, to creating the materials that manufacturing would use, to producing the orders. The team included the supervisor of the off-set printing and die cutting departments, the manager of sales, a representative from order entry, an operator from the screen printing department, a member from accounting and a production artist.
Because this company had already completed training for all employees in Lean Six Sigma techniques, they started with a round of brainstorming to create an initial hypothesis about the issues that needed to be improved upon, in tandem with reducing errors. The list included:
- Wasteful double-checking of orders versus getting it right the first time.
- Rework of information.
- Reducing the number of process steps, hand-offs and physical walking in the order entry process.
- Jobs getting “stuck in the boss’s office” – waiting for approvals.
- Jobs waiting for files, samples, data entry, orders, and answers from those we need to consult with, including the customer, managers and key subject matter experts.
- Processes are not linear – wasted moves, hand-off delays.
- Searching/looking for stuff and information i.e., “digging orders out of the costing stacks.”
- Layout of the facility drives wastes.
- Hunting down managers for questions.
- Making copies for job packets – dies, artwork and information that does not fit on the job jackets – not even sure why some copies are made!
- Taping pages to job jacket, cutting/pasting/stapling stuff to the job packet equals wasted effort and lack of standardization.
- Under-utilized human resources – not trusting it’s right the first time equals a lot of effort in double, triple checking.
- Shipping labels printed in the office, then carried out to the shipping area.
- Lack of standardized information equals confusion, guessing and mistakes.
- Different bosses want the same things done differently.
- Too much of the information in our job packets is subject to interpretation – meaning different people would come to different conclusions on what to do from a single order.
- Too many latent (hidden) defects in our paperwork makes for a “culture of checking.”
- Too many hands touching the orders – need to stream-line the process.
- Lack of effective training and people not understanding turns into mistakes and delays.
- Under-utilized technology – wasted effort because people don’t know how or won’t use what we have.
- Redundant data entry and manual writing of the same information over and over.
- The sales/order entry process is not effective in doing it right the first time leading to many mistakes, rework, confusion, delays, etc.
If any of these problems sound like what happens in your organization – you are not alone. My experience in hundreds of businesses of all types is that we tend to shy away from using Lean Six Sigma techniques in office settings because we are not “making something” – a product. This paradigm is just plain wrong – as this team was able to prove.
First Step – Getting Some Data to Work From
As is the case in all office processes there were no measures of the process in place to work from as a baseline. What the team did was construct a type of tally sheet that listed all the typical process steps for processing orders. These were generated at the initial phone call and “rode along with” the orders during their entire life-cycle into the art department. Over a two-week period, they tracked and captured:
- All the steps to process including rework.
- Cycle time (human touch time).
- Delay (delta) time – an order is waiting to be worked on for some reason.
- Number of calls to close the sale and gather required information to process the order.
- Batch versus single order moves.
- Other anecdotal information.
The team reviewed this information and came up with a method to convert it into a usable form. All the order tracking documents were decomposed into the data needed and combined into a single worksheet/chart. From this they were able to determine:
- Typical workflow steps up through President approval, not including Art and beyond (14 today).
- Too many operation steps that create rework – at least five steps in the current state process.
- An estimate of the number of orders requiring proofs and the resulting work and delay time was derived.
- The low, high and average (mean) cycle times for orders was calculated – the cycle time (“human time”) was about 52 minutes per order on average to get them to the art department.
- The number and time of delays (deltas) was calculated with a low, high and mean – 14 delays, between 17 and thousands of minutes in duration for each delay.
The team realized that their cycle efficiency for the process was very poor – and that taking time out of the process was important to satisfy the company’s ongoing objective to reduce lead time. The team created a VSM to provide a visualization of the workflow with this data and drew-in the rework loops as seen here:
Armed with all this information, a number of people were asked to participate in brain-storming ideas to improve the process. Some ideas developed and selected for quick implementation included:
- Collapse steps for standard/stock/re-orders to a very few steps (hand-offs) i.e., standard, high-volume repeat orders. This decreased double/triple checks, thus cutting delay time by 50%, and took the President and sales manager out of the loop (they were not looking at them anyway!).
- Standardizing the process by creating checklists for incorporating the methods for taking or proofing orders – and creating standard visual work so that orders are processed the SAME way by everyone, every time.
- Revise work orders to be more consistent, easy to read, not “subject to interpretation” – so that there is only one way to interpret the information.
- Reducing rework by identifying the root causes for rework and eliminating them.
- Increase quality and effectiveness of training for sales reps who take orders.
- Create better flow of the work by implementing an “office cell” for repetitive tasks.
- Charter a project to collect root cause data on mistakes for later Pareto analysis and improvement.
Those All-important Measures of Success
The team then established their initial goals and objectives for their improvement project as:
- Increase first time capability, reduce rework and reduce returned orders by 33% in 60 days.
- Reduce “delta” (delay) time and handoffs 33% in 60 days.
- Reduce human effort by 20% (by reducing waste) in 60 days.
- Gather the data for mistake root causes – and reduce mistakes 50% over six months.
A detailed action plan was developed with clear accountability and deadlines for action. In the second of this two-part series of articles I will detail an important implementation effort of this team – finding and attacking the root causes for mistakes in this process – one of the main charters of the team being an objective to reduce re-prints in manufacturing.
Yet another round of data collection
In their initial VSM effort, they highlighted where they thought the errors were occurring – but had to rely on anecdotal information to create hypotheses. It was agreed that much more quantifiable data was needed so that the true root causes could be isolated for corrective actions with the right process metrics. After some brainstorming, they decided to set up a data collection method to capture root cause information for each order that was returned from any downstream step for corrections.
A spreadsheet was created to tabulate the following information for each return for corrections:
- Date of the discovery of the problem.
- The identity of the sales representative who processed the original order.
- Which of their defined markets this was for (to identify any correlations).
- A description of the error – this was a short-hand code, such as “Numbering,” “Layout” or “Finished Size.”
- Who the error was made by – as there were hand-offs for some process steps that the sales representative does not do.
- Problem description detail – specifically what was wrong.
- A notation – about whether or not the information was provided correctly on the order form in the first place. It turns out that there were four possible answers which were important to learn – including; No, Yes, Not Applicable and Not Clear.
Over a two-month period 137 observations were added to the worksheet from which the team was able to perform some initial Pareto analysis (this is also known as the 80-20 rule, or the maxim that 20% of causes account for 80% of the effects).
Uncovering true root causes is the mission in this kind of improvement project. The team took great pains to make sure that everyone involved in the data collection understood this was not a witch hunt. To avoid bias in collecting the information the team members took the time to examine each reported incident carefully and perform a root cause investigation within 24 to 48 hours before the annotations were finalized in the spreadsheet. People who participated were thanked and those who assisted were singled-out for a special “thanks” from management.
They tracked the error rate on new orders entered over the two-month period and determined their “first time right” rate as 71%. Their short-term objective was to improve this by one-third – or move it to at least 80% right in the first 60 days after implementing countermeasures.