8 replies to this topic

[C_Custer]

I'm wondering what percentage accuracy you should get on a mock recall. Of course I'd love to get 100% but doesn't seem possibly at the moment. Nothing is mentioned in the SQF code so I'm wondering what everyone else uses as reference to verify your system is good enough.

Thanks

[olenazh]

Are you asking about recovery %? Then, ISO 22000:2018 doesn't specify it as well as CFIA. We've established our recovery range 99.5% - 100%

[The Food Scientist]

If you don't get 100%, you have to explain why. But I always put 95-100. 

 

That's what I do. Works everytime.

Edited by The Food Scientist, 08 June 2020 - 05:46 PM.

[SQFconsultant]

You are right, it doesn't say the percentage.

 

We shoot for 99-100% however.

 

If I was doing a mock recall and came in at anything under 99% I'd be looking very closely at why that is, go through identifying the issues and correcting them.

 

It is easy to put a .5 or 1% into waste, samples, etc. - we would be hard pressed to explain to an Auditor that we could not account fully.

 

Keep in mind the Auditor may have you conduct a mock recall at time of the on-site certification audit.

[C_Custer]

Thanks everyone for your feedback. We've always done a mock recall on packaged lot numbers and easily gotten 100% every time. I decided to do a complete recall thinking what if our supplier initiated the recall. So we went from raw all the way through finished product to customer. We're now finding issues with our counts. We have another meeting this afternoon to figure out corrective actions and dig deeper.

[TimG]

In the sugar industry we put on paper 95-105% (yes, 105%). Sugar being bulk involved quite quite a bit of floor sweep (which was tracked and went to rework). 98-102% was a typical range found on a trace exercise and anything outside of this prompted a further discussion to find out why, but we would still pass the trace exercise. Each new SQF auditor after learning the process had no issues with the 95-105% as the threshold for fail.

I guess that story is just there to tell you it really depends on your process and the measurement of accuracy achievable.

[MsMars]

Thanks everyone for your feedback. We've always done a mock recall on packaged lot numbers and easily gotten 100% every time. I decided to do a complete recall thinking what if our supplier initiated the recall. So we went from raw all the way through finished product to customer. We're now finding issues with our counts. We have another meeting this afternoon to figure out corrective actions and dig deeper.

 

It's always good to do the complete recall from ingredients to finished products that way you've covered all your bases in one go.  

[Wine_QT]

This is interesting as we've had a horrible auditor (when they ask what does the B,C,P letters mean in a HACCP audit... really.. really...) and he stated on our last audit we had 99.8% on a mock recall and we could speak to it and he said well he considered that a fail. So frusterating. This wasn't a SQF audit.

[Charles.C]

Possible that more than one interpretation of "recovery" is occurring in this thread.

 

This detailed example seems like a  useful model although it omits timing criteria -

 

https://www.ifsqn.co...ort/#entry47832

 

5% negative deviation being regarded as acceptable seems questionable IMO. Especially if it cannot be plausibly accounted for.

 

A 5% positive deviation seems even more questionable as far as data reliability is concerned ?

 

PS - in respect to interpretation -

 

 AIB the mock (recall) myth.pdf   287.56KB   134 downloads