# RFID Arena

## Impact of Item Level RFID into Inventory Accuracy

This paper discusses the impact of item level RFID in inventory accuracy. The paper uses a simple mathematical approach for simulating the effects. The results are shown visually as well as in numeric data.

INVENTORY ACCURACY DETERIORATION

Understanding inventory accuracy deterioration is a starting point for evaluating impacts of RFID counts. The best way to find out how accuracy deteriorates, would be to use real measurements and tests. Also, in most real life cases responsible persons in stores can give good estimates and explanations from their experience. For generic analysis a mathematical model is useful, with mathematical model it is easy to analyse effects of various parameters into end result.

INVENTORY ACCURACY DETERIORATION - MODEL AND EXAMPLE

As an example we use a store that has 1000 items in 1000 locations. An item file describes where each item is. In this item file there are 1000 data points, each of them is either correct or incorrect.

In the store may occur 5 different types of incidents that deteriorate the item file accuracy. These incidents have different probabilities of occurrence, i.e. they occur at various intervals. The incidents occur for 1000 item population as follows:

• A: 3 times per 1 day
• B: 2 times per 1 day
• C: 1 time per 1 day
• D: 1 time per 2 days
• E: 1 time per 3 days

In the beginning (day 0) the item file has 1000 correct lines. Based on the occurrence rates above there are 3+2+1+0.5+0.3 incidents = 6.8 incidents during the first day. After the first day the item file has 993.2 correct points and 6.8 incorrect ones.

In the example only incidents deteriorating the inventory accuracy are being used. This model includes also a possibility to model incidents improving accuracy - such as store personnel placing misplaced items correcly, and incidents effecting both ways - such as a customer placing items randomly. For understanding and analyzing the item file accuracy development in a generic level it is better to focus on some clear incident types.

The model used in this paper is a simple excel chart that is used for calculating item file accuracy based on the parameters described above after each day. For simplicity only the accuracy end of each day is calculated, as if all incidents would occur at a single point in time at the end of each day. In the model also the item quantity is static - all items sold are replaced with same items during the day. In the model incidents and developments happening during the day are not factored in.

BASE LINE - 100% ACCURATE INVENTORY COUNT ANNUALLY

With the model and parameters described above the item file deteriorates as described in the graph below.

Picture 1: Item file accuracy over 360 day period without inventory counts.

For over 1 year period the average accuracy with these parameters is 37.1 %.

As we can see the item file accuracy deteriorates faster, when the file is correct. Over a long time period the asymptothically approaches 0.

QUARTERLY 100% ACCURATE INVENTORY COUNTS

When inventory counts are made quarterly with 100% accuracy, the Item file starts deteriorating after each count and in the count returns to 100% accuracy. The result is shown in a picture below.

Picture 2: Inventory accuracy with quarterly counts.

In this case average annual accuracy is 73.5%.

FULL WEEKLY RFID COUNT WITH 97% ACCURACY

In the first RFID based scenario counts are done weekly and the accuracy of the count is 97%. Results are described in a graph below.

Picture 3: Inventory accuracy with weekly full counts using RFID.

In this case annual average accuracy is 94.5%.

From the graph the deterioration and correction of the accuracy level up to 97% can be seen. In this case the first count actually decreases item file accuracy since in a week it has not deteriorated to 97%.

The Nordic ID Merlin UHF RFID Cross Dipole is the best suited RFID reader for this type of use scenario. It has a long battery life making full counts efficient. Nordic ID Merlin UHF RFID Cross Dipole  has a RFID performance of top level making the inventory accuracy high.

10% OF THE ITEMS ARE COUNTED DAILY WITH 93% ACCURACY

The second RFID based scenario is to use more frequent partial counts. In this case item file accuracy reaches a minimum level. For example if 10% of the items are counted daily with 93% accuracy the results are as shown in the graph below.

Picture 4: Inventory accuracy with daily partial RFID counts

The average annual accuracy is 87.9% and over a long period of time the accuracy settles to 87.6%.

The Nordic ID Morphic UHF RFID Cross Dipole is an ideal reader for this kind of scenario. It is small sized, light weight and thus easy to carry around. Store personnel can carry the device and perform inventory counts for individual shelves, rounders, display tables etc. when the work load allows.

COMBINATION OF PARTIAL DAILY RFID COUNTS AND MONTHLY FULL RFID COUNTS

The third RFID based scenario analysis combines partial and full counts. In the example 10% of the items are counted daily with 93% accuracy and a full count with 97% accuracy is performed monthly. The results are shown in the graph below.

Picture 5: Inventory accuracy when partial daily counts and monthly full counts are combined.

In this scenario average annual inventory accuracy is 90.2%.

CONCLUSIONS

Mathematical model is a usefull tool for evaluating and visualizing item file accuracy with various RFID based count scenarios - it can be used as a simple way to evaluate the effect of RFID count accuracy and RFID count frequency.

Item level RFID enables frequent counts and thus the inventory accuracy improves. In the example case average annual inventory accuracy is

• 37.1% with annual count
• 73.5% with quarterly counts
• 94.5% with weekly full counts with 97% accuracy (Nordic ID Merlin)
• 87.9% with daily 10% partial counts with 93% accuracy (Nordic ID Morphic)
• 90.2% with combining 10% partial daily counts with 93% accuracy with 97% accurate monthly full count

Using the item level RFID has clear benefits in the example case. Both full periodical counts and frequent partial counts are usable. Also a mixed model is feasible. Selection between these models can be based on the other store processes.

Per request Nordic ID can produce inventory accuracy simulations  with different parameters such as various RFID count intervals and accuracies, various deterioration parameters and various starting positions.

### 3 comments on “Impact of Item Level RFID into Inventory Accuracy”

1. murat
Posted 16 May 2016 at 15:26:18

what is the reference list ? I want to use it in my homework. Can you help me

2. Posted 02 January 2018 at 15:58:08

Hannover does not hold many events but this Cebit event was definitely welcome here. Many people from the Rush Essay were at this event and every one there had loads of fun. I would love to go to an event in Hannover again.

3. Posted 28 May 2018 at 09:54:00

I really enjoyed the top quality info you present to your visitors, will be back again frequently to check up on new posts. It is happened that I truly cherish
perusing this sort of articles. I don't have a clue, however I adore it.