Salman Quadri runs Taskeater’s Quality Assurance team, which was introduced late 2015 to work across departments in the company. Introducing a formal QA process was a major milestone for the company, however the team’s evolution over the year has been an organisational learning experience and has required the organisation to constantly adapt to new ways of working.
There are two main types of human error that can occur in business process outsourcing (BPO):
- Errors due to insufficient training
- Errors due to workflow fatigue
Errors due to insufficient training
These are most likely to occur during the onboarding and training period, during which the team’s line manager checks 80-100% of the completed work in order to assess the how far along the learning curve the team’s analysts are. Only when there is confidence from both the line manager and the client that the knowledge has been successfully transferred does Salman step in to begin assessing the right method for mitigating against the second type of error: workflow fatigue.
Errors due to workflow fatigue
Most of us at some point have experienced workflow fatigue, which typically creeps in over time as we begin feeling overly confident in our ability to manage work we feel comfortable with. We tend to look for shortcuts and rely on muscle memory to navigate workflow rather than thinking through the fundamentals of why certain steps are being taken. To mitigate against this, Salman typically sets up a system for a Quality Assurance Associate to check 5-10% of work being delivered. With some clients acceptable error rates are established (balanced with speed of delivery), while with others quality is more subjective. This acts as an early warning system, helping to proactively detect early signs of workflow fatigue. QA Associates report back using standardized spreadsheets which both Salman and the line manager can look through.
Methods for checking output
Most importantly, Salman must assess what method of checking output must be employed in each case. The three main ones are:
- Post-delivery checks: About half of all assignments can be checked by a QA Associate after the work has been processed as the finished data is stored separate from the raw data.
- Live checking: Some assignments deal with live data processing which is all done in an online back-end portal. In such cases, QA Associates must spend an allotted period of time watching the analyst perform the task live and take note of any deviations from previous workflow.
- Auto-checks: In some cases, especially if working in spreadsheets, additional formulas can be added to automatically check for deviations. An example of this is financial data transcription as certain numbers can be cross-checked through formulas that flag potential errors.
Thinking about QA is often a mathematical exercise as well, as Salman is essentially tasked with reducing the probability of error. It is important to assess what areas of work are most prone to errors and craft processing in ways that mitigate against them. For example, when populating spreadsheets with data, certain information can be copy-pasted directly into cells while other cells are must be typed manually. The latter is more prone to errors and therefore requires more attention from QA Associates. Furthermore, often some cells can be auto-checked using formulas while others must then be checked post-delivery.
Salman has had an interesting journey getting to know virtually all of the company’s assignments at a micro level. Today his role is to educate line managers and analysts in QA techniques, and the thinking behind QA probabilities. With the whole organisation using the same mindset of delivery working hand-in-hand with QA, and information being looped back in a knowledge transfer cycle, Salman hopes to continue reinforcing the QA team’s role as a major selling point in the BPO industry.