How to benchmark lab turnaround times?

There are many ways a lab can provide you data about turnaround times within the lab (i.e. Time taken from point of arrival at lab to point of the report being loaded online or emailed). Some labs measure in days, or average number of days, which although the simplest to understand it says nothing about the lab performance unless you only ever have one sample type with one suite and never any variations. When presenting the data graphically the scales are often presented with lots of days to make the numbers look small and often put next to the courier transit times (i.e. time from taking sample to arrive at lab) which will usually be high unless the client has perfected their own sampling process to ship on day of sampling.

An example of a traditional presentation of data is something like the below.

Many labs use this style, but as a client trying to compare which lab to pick to start sampling with this data can be meaningless if not put in context. I.e. a lab that 90% of their work is innovative research type analysis will always look bad on that scale, whilst a lab that does the most basic test suite will always look fantastic. For a like for like suite the data may be completely different.

Another common style is showing as a percentage by suite type how many achieved their turnaround or not. This was for many years my method of choice of presenting the data, however, this still did not help the customer because take the scenario there are two labs doing the same suite for a client:

Lab 1: Turnaround 90% on time and 10% late.

Lab 2: Turnaround 85% on time and 15% late.

Assuming all things equal you would naturally look at Lab 1 and ignore Lab 2 going forward. However, now lets say the desired turnaround was 2 working days.

Lab 1: Turnaround 90% of the time in 2 days, but all loaded a few seconds before deadline and the late samples because already late were left an extra couple days and then reported.

Lab 2: Turnaround 85% on time, but 75% were done in first day and 25% the lab thought should have some additional retesting to confirm some expensive corrective actions and this meant that 10% were reported the following day (i.e. still within turnaround), but the remaining 15% were all 4 hours late because of that confirming.

With those types of stats there is no incentive once late to try get them out as soon as possible, and there is also no incentive to try impress the client and get some of the work out earlier than expected.

That is why from my years of experience in the industry I introduced the Learnoilanalysis Perfection Index (LPI) for turnaround times.

In an LPI you start with the turnaround being 100. If all samples go out as promised with no late samples then the score is 100. Then values are added or subtracted accordingly.

A sample that is late by <24 hours, e.g. by a few hours gets a score of -1.

A sample that is even later i.e. by >24 hrs gets a score of -2

However, for every sample that is >24 hours early the score is +1

In this scenario being marginally late occasionally is not penalised as long as the lab make it up by being early on other occasions. There is still an incentive on late samples to make sure they don’t go very late to avoid the score becoming twice as bad. Being more than a day early does not give any additional incentive meaning if the lab let you down and are >24 hrs late then they must make it up to you twice.

Some Example LPIs are shown below (left – lab starting to under-deliver on promises), middle usually delivering what promise and right a lab that is not only meeting their promises but exceeding them. Note between the middle and right graphs there is only 6% difference in traditional turnaround of on time or not, so this Index allows the reader to more accurately determine how the lab is performing and gives the lab the opportunity to impress their clients by exceeding their expectations.

A lot can be done to improve the total turnaround times by addressing non-lab associated delays by streamlining your processes (see “I feel the reports are taking a long time to return” previous section). However, you can also help the lab in the following ways.

  • Plan your sampling.
    • Make sure the lab knows your planned maintenance scheduleIf a sample is going to be urgent take a reference of the bottle / form, ship on a priority delivery courier and let the lab know the bottle number and courier tracking number so they can expedite that urgent sample for you.
    • Sample well ahead of your planned shutdown or maintenance. If you know in 1 month you will be shutting down your plant for 2 weeks don’t leave it until a couple days before to take the samples, plan in advance so you have time for the routine sampling + further testing on identified problem machinery.
    • Pre-Register – Rather than fill in a paper form which will need processing by the lab before some tests can be performed, why not register the sample using your labs online registration systems, so the testing can start the moment the sample arrives with no delay on processing the paperwork. If you really must use paperwork then use the lab paper forms rather than compliment slips, yellow post it notes or marker pen or white liquid paper on the lid.
    • Use the correct container – Every single day the lab receives samples in coke-a-cola, water bottles, specimen pots by clients who have run out of bottles etc. Not only are these bottles not suitable for transporting oils, but they also cause delays in extra subsampling must be done to process them, plus since these are often unpaid it can also give both yours and our accountants headaches. Go to your stores now and if you do not have enough samples to take everything you need for the next month and still have some spares for any adhoc or repeat samples then you need to order in more stock.
    • Match the criticality of your sampling reflect your schedule – We often see clients batch up an entire year of sampling into a couple days to send off and then express all of them are urgent. There are many reasons people do this such as waiting until just before an audit to do sampling, sampling in advance of a shutdown, convenience or to save on transport costs. In each case as mentioned above, letting the lab know in advance what you are planning to do, sampling the batch before the planned shutdown and pre-registering the samples. You also might want to ask yourself if the samples are so critical that they require urgent status then why are you not sampling some of these assets more frequently so you can have your maintenance planned well in advance rather than being so reactive to a single annual check.
  • Look at the bottle before sending it
    • If you can see a layer of water, pieces of metal, plastic, rubber, sand or grit at the bottom of your sample ask yourself do you really believe that is what is representative of the system you just sampled. If the answer is no, then take another sample first before sending off. There is nothing worse for a client urgently requiring a report to be told they need to resample again when the issue could have been spotted at point of sampling.

If the oil volume is quite small and easily changed at minimal cost some corrective actions can be actioned before sending to a lab. For instance, if you see the oil is excessively cloudy, emulsified and has some free water and have confirmed it was not a bad sample then you may choose to do the oil change or filter as the diagnostic advice will usually include an this type of advice as part of the corrective actions. You can then after running the oil for a few hours at operating temperature send a new sample off as a baseline to follow-up the contaminated sample. We still recommend sending the contaminated sample too as this will ensure the diagnostician has a complete history of the equipment and can also help analyse what damage was done by the contamination and potential causes too. You can always add in the sample comments when submitting samples or submit this information through your lab portal database.

Leave a Reply

Your email address will not be published. Required fields are marked *