Prekitting

[Chard]

What is the best method for prekitting, MDB or Dex?

We have a wide mix of equipment ranging from upgraded AP 113's, LCMs (Stock), Studios, Nationals 157s, 167s, AMS, some USI stuff we aquired along the way.   

With such a large mix, we've got stuff that'll dex and stuff that won't.  We've got machines that send good MDB data and a lot that won't.  
Regardless, we're moving to prekitting, but the process is tedious at best.

I'm growing concerned we're going to have to test each machine one at a time to figure out the mapping if we use MDB.  That may break the process.  For those who have done prekitting using data, how did you do it?  What did you learn?  Which method was best.

Any help is greatly appreciated.

 

[lacanteen]

DEX is most accurate then fed to Lightspeed. 

[Chard]

What do you mean by most accurate.  Does MDB feed false reads or is it simply slower to update?

[putrevus]

20 hours ago, lacanteen said:

DEX is most accurate then fed to Lightspeed. 

So how do you prekitt using Dex if you are taking the Dex readings while filling the machine which is near empty.

[AZVendor]

You don't do prekitting on location. You read your data remotely before loading your truck and each location gets just what they sold and that is all you take in to the location.

[lacanteen]

What we pick today is put in the machine tomorrow. DEX telemetry is sent 2 times a day between 2:30 and 4;30 (AM & PM). Orders are locked at 5AM. The system takes the exact item sales per machine then adds 1 ADS (Average Daily Sales). So if the ADS for Coke is 6 and 36 were sold, we will send 42. Our trucks are loaded in the late morning & afternoon. Most drivers take their trucks home and start work in the morning. Most start between 3 AM and 6 AM. 

Soft drinks are picked in aggregate, in other words, the total of each item for all scheduled machines is pulled and rounded to a 6-pack. Snacks are picked in individual totes. The driver's handheld tells him what drinks he needs when he arrives at the stop. 

[putrevus]

15 minutes ago, lacanteen said:

What we pick today is put in the machine tomorrow. DEX telemetry is sent 2 times a day between 2:30 and 4;30 (AM & PM). Orders are locked at 5AM. The system takes the exact item sales per machine then adds 1 ADS (Average Daily Sales). So if the ADS for Coke is 6 and 36 were sold, we will send 42. Our trucks are loaded in the late morning & afternoon. Most drivers take their trucks home and start work in the morning. Most start between 3 AM and 6 AM. 

Soft drinks are picked in aggregate, in other words, the total of each item for all scheduled machines is pulled and rounded to a 6-pack. Snacks are picked in individual totes. The driver's handheld tells him what drinks he needs when he arrives at the stop. 

Thank you for your help, is that DEX telemetry sent by USA tech report or any other service is there to do that? How do you get DEX reading for machines which are not having Card readers.

Edited December 16, 2019 by putrevus

[lacanteen]

DEX telemetry sent by USAT to Streamware. Streamware creates the machine "service orders" and feeds the data to Lightspeed. 

[putrevus]

1 minute ago, lacanteen said:

DEX telemetry sent by USAT to Streamware. Streamware creates the machine "service orders" and feeds the data to Lightspeed. 

Thanks again,  this works for machines with only card readers,The machines without card readers still have to be done prekitting manually?? 

[lacanteen]

Machines with no card reader are estimated using ADS. 

[putrevus]

1 minute ago, lacanteen said:

Machines with no card reader are estimated using ADS. 

Thanks  a lot this is what I need, ADS,

where can I get ADS thanks again for your help.

[AngryChris]

You need a VMS (vending management software) such as par level, cantaloupe, vendsoft, etc..  I use vendsoft.

[lacanteen]

It took us months to settle on a formula for the ADS algorithm. I don't know where it is now, but back when we started this 6 years ago, we took sales over 20 weeks but weighed 50% of the number to be the average of the last 4 weeks. That helps account for seasonal spikes in retail, schools, and hotels. You also need to take into account if a location is open on weekends or not. 

You have to play with the algorithm to see how it fits your business. It took us 4 months of experimenting to get it close. I'm currently encouraging the use of fractional ADS depending on what time of day the machine is scheduled to be filled. A high ADS machine would get too much product if filled early. 

[putrevus]

4 minutes ago, AngryChris said:

You need a VMS (vending management software) such as par level, cantaloupe, vendsoft, etc..  I use vendsoft.

I wanted to use Parlevel for offline machines but was not really sure about it.Since all our card readers are with USA tech.

Edited December 16, 2019 by putrevus

[Southeast Treats]

23 hours ago, putrevus said:

I wanted to use Parlevel for offline machines but was not really sure about it.Since all our card readers are with USA tech.

That is exactly what I have, ParLevel VMS but USAT telemetry.  About half my machines are still offline, and ParLevel does a good job on the predictive prekitting as long as you have good data going in.  PL lets you set the number of offline past services to be averaged to get the offline prekitting, anywhere between 1 and 10 I believe.  I use 3 services as my average point and it works well.  I use the app and record offline services on my smart phone; my driver uses a tablet.  Both ways work well.

[TKK]

Good luck, i spend hundreds of hours and thousands and you need months of data and training to tour enployees to launch it. My problem was when ccs go offline and it messes up your prekit. May show 0 sales then your guy takes 0 items. Then tou have to manually put items and modify...

[Chard]

So far I've pulled kits estimating sales from previous services.  Say par level is 6, we just send out 6 bags because the row typically sells out,   you find out 5 sold, so we bring the etxra bag back to the warehouse and pack it in a kit for the next day and move on.

Teletrmy will certainly improve estimation, but this has been a great start.  

So far, it appears to save a lot of time.

[AngryChris]

33 minutes ago, Chard said:

So far I've pulled kits estimating sales from previous services.  Say par level is 6, we just send out 6 bags because the row typically sells out,   you find out 5 sold, so we bring the etxra bag back to the warehouse and pack it in a kit for the next day and move on.

Teletrmy will certainly improve estimation, but this has been a great start.  

So far, it appears to save a lot of time.

I'm assuming that was just an example but.. if you normally sell out 6 by the time you go to service, you'd want to set your par higher (at least 8 or so, if possible).  That way, you're less likely to get calls/complaints about being sold out of an item.  That may not work for all things but Ideally, I want everything to have 1 or 2 items in it.  Things like crackers which last months, if I sell at least half of the entire row between service cycles, get maxed out in the selection to minimize lost profits due to being sold-out.  Pastries are kept barely above what would sell-out and sometimes I put slower selling (longer date) items behind them so they don't look so bad right before I service it.

[lacanteen]

Selections that are continually sold out cost you sales. We have an analyst that not only merchandises he specifically looks for "outs" and "zero turns". (empty and no sales)

A computer is only as smart as the data. Data is only useful if you use it.