Quick Wins – Five steps that can quickly improve a process
A small improvement can go a long way. Not all process improvements have to be fundamental overhauls or redesigns. There are a number of things that I have found can be done quickly and immediately to improve a process or a report.
I swear by checklists. A simple list of what needs to be done during the process, when and by who. This could be to check that the monthly data has been downloaded on the right day, or that each spreadsheet page has been updated.
A checklist has two effects. Firstly, it gives the discipline to remind people what to do – following and ticking off a list makes it easy to remember everything. Secondly it gives visibility that everything has indeed been done – it is a completeness check for managers, stakeholders, and above all, auditors, to have confidence in the process.
I came across a process that required the standing data mapping of accounts to cost centres to be updated each month, the mapping being used in subsequent reporting processes. I found that the update had in fact not been done for well over a year. This is a common issue – standing data can quickly get out of date unless it is regularly updated. I introduced a simple tick checklist to show when it had last been updated, making the data that was used in the subsequent reporting processes much more reliable.
2. Map the controls
All parts of a process should be controlled, in one way or another, to make sure the process is being operated correctly. The integrity of the whole process is only as good as the sum of its parts.
And a picture can paint a thousand words. So, I find it incredibly helpful first off to quickly sketch the process. Then I can overlay the controls that are being operated, and easily see if any parts of the process are uncontrolled.
Here is an example of a process I quickly mapped, which showed that one part of the process was exposed as uncontrolled:
Controls need to ensure that the correct data is processed in the correct manner. From this quick map, I could see that controls were in place to ensure the correct files were 1) extracted from the source commissions and ledger systems, and 2) loaded into the combined transactions spreadsheet. But the process to then cleanse that data was totally uncontrolled (a possible control would be an exception report of cleansed data – original vs. amended values). This, in effect, compromised controls over subsequent parts of the process – loading the cleansed data into the FX reporting system and then extracting the month end reports and BI. There could be no assurance that the end reported information was correct because all parts of the process to produce that information were not properly controlled.
3. Expected vs. Actuals
Another quick action is to insert a sense-check into the process or report. Sense-checking is seeing that data looks right. It can be applied to both the source inputs to the process and to the reported outputs from the process. Is the data what is expected? Is it within expected ranges?
- Values – set a high/low limit check on values – perhaps the total for a report, or for particular cost centres. The value between one month and the next should be fairly consistent. Putting in a quick check whether the current value is consistent with the historical range will pick up on any wildly wrong data.
- Dates – set a check that all transactions occurred within the last month, for example. Or that the source file being imported was generated on the correct day – i.e. the prior month’s file has not been imported in error.
- Volumes – check that the number of lines of data is consistent with what is expected. Or that the file size of any source file is ok. Any wide variations in the volume of data being processed may suggest a problem.
4. Control totals
Another quick, but essential, check to have in place is whether the data output is compatible with the data input. Has all the data been processed correctly, or has some been missed, or duplicated? This can be checked by creating control totals.
For example, the values of the data input can be totalled, and this total value then compared to the total of the information reported. They should, apart from known exclusions, be the same.
Any differences between the totals input and output should then be analysed.
There are quick wins to be had in automating at least some of the process. Data and files can be linked to, rather than input manually, for example. As well as speeding up the process itself, this also adds an element of control because manual input can always lead to errors. An automatic link to data can be established, with a sense-check control to see that the linked data is valid perhaps.
So those are my five quick wins! They are what I look to implement straight away from day 1 of dealing with a process. None of them are that difficult, but they require standing back from the process to see where these are missing from and where they can be best placed.