CPG-Retail

The Why and How of Price Detection Through Retail Image Recognition

Ankit Singh
January 4, 2021
3
mins read

Price detection and its compliance is an important part when it comes to achieving a perfect store programme. Brands are looking towards retail image recognition solutions for the same. Monitoring price compliance through price detection is an important feature of our AI retail image recognition solution. In this blog, we discuss why price detection is important and provide a bird’s eye view of how the technology works.

WHY SHOULD CPG COMPANIES MONITOR DISPLAY PRICES? 

The importance of price display monitoring for Consumer Packaged Goods (CPG) companies mainly comes from  high instances where incorrect prices are displayed to the user than what was intended. Some such instances that lead to it are - 

  1. When retailers don’t follow the guided price range.
  2. A missing price display.
  3. An incorrect location of price display.
  4. Promotions (like discounts and combo pack prices) may not be reflected in the price display.
  5. Price display does not reflect the changed pricing.

1. WHY DO INCORRECT PRICES GET DISPLAYED?

The reasons for the above situation could be myriad.

Retailers may cut or raise prices Another reason could be that the retailer might not have an updated database for changed pricing.

Not only this, when the customers move around the store, they may pick up a product and place it somewhere else. This disturbs product placement and consequently, the price display that has been allotted to that product.

Often retail representatives are responsible for handling a lot of products. They have to put in order several products and keep it in sync with the Point of Sale Material (POSM). It's a big task to undertake and could sometimes lead to erroneous execution. All this could potentially lead to incorrectly displayed prices.


2. THE IMPACT OF INCORRECTLY DISPLAYED PRICES

Some retailers may aggressively slash or raise prices. If the prices are raised then there is a loss of sales. If prices are slashed then the company loses revenue. Either of these scenarios do not align with the company's strategy.

Also, a not updated database could lead to inconsistent pricing across different outlets. This would be at odds with your brand strategy of giving a uniform customer experience. This unplanned inconsistency in pricing could also hurt your retailer relationships.

Then there’s the instance of an incorrect price display. For instance, let's say a customer enters a store to buy a shampoo belonging to Brand ‘A’. But when they reach the aisle, they encounter a wrong price display which has been attributed to it. The unsuspecting customer may just assume it's the price of their shampoo and decide to buy its cheaper alternative, say one that belongs to a brand called ‘B’. 

After a few days, this customer requires a conditioner. It makes more sense for them to purchase the conditioner of Brand B which compliments their shampoo. Not to mention that people form a routine and stick to it, which means they are now friends with Brand B.

This causes a direct impact on sales and the strategy that the brand intends to follow in the market. It is also a cause of poor customer experience, This could make sales loss permanent and may also extend to other categories of the brand.

In the same manner, an incorrectly displayed price would negatively impact sales when promotions are being offered by the brand. During promotions, rules for POSM have to be complied with, especially correct price display. A wrong price display here would essentially thwart the entire reason for the promotion.

DIFFERENT WAYS OF MONITORING PRICE:

Different methods of monitoring price tag detection by sales reps, third party audits, retailers and use of AI image recognition
                     

Key Performance Indicators (KPIs) like Planogram compliance, POSM compliance are aided, and their standards improved with the help of price detection feature. Also, CPG companies keep in mind foremost, the issues that a wrong price display can lead to. As a result, they conduct price monitoring exercises as a part of their routine store visits. These visits can be carried out by:

1. Third party Audit: Herein the FMCG in order to implement their perfect stores, creates a set of standards with the help of KPIs. Then they hire an independent auditing company to visit their stores and see if those standards are being implemented.

2. Self reporting by field reps: In this case, the company uses its own field reps who routinely visit stores and they see to it that compliance is established. Wherever it is not, the measured KPIs give them direction to proceed and try to create a perfect store condition for their brand.

3. Retail Partners : Often retail partners are encouraged to gather information regarding price detection data. The reports they generate are used by the brands.

These methods are mostly manual. And us humans? We are subject to bias. 

Self reporting by company sales rep’s is essentially a conflict of interest. They may not relay the ground reality to ensure their sales target is met for the month. 

And as for retailers, especially in general trade, there’s an absence of a standardised system that efficiently catalogues brand sales’ in different categories. Either way, their perspective when storing information is the store’s sales and not brand health.

This has led to the advent of a new player : AI powered image recognition and object detection solution that strives for retail compliance in an objective and measured manner. In fact, as per Gartner report, Image recognition technology can increase sales force productivity, improve shelf condition insights and help drive incremental sales. 

IMAGE RECOGNITION FOR PRICE DETECTION:

This solution involves a software that uses image recognition to identify the brand and category and object detection technology for identifying SKUs. 

Herein , the sales reps or the third party auditors use cameras or mobiles to capture pictures of the shelves. Then these pictures are sent to the cloud server where the AI processes it. It detects the SKU and then calculates the KPIs associated with it. They may include out of stock, share of shelf and price detection.

Understanding the technology in more detail:

Understanding the technology behind price detection by AI solution for retail image recognition

The AI is trained to recognise the brand’s SKUs with the help of images sourced from the FMCG. When deployed on field, it follows the following process in order to perform price detection:

  • Step 1 - The AI firstly detects all SKUs present in the picture
  • Step 2 - The AI then detects the shelves present
  • Step 3 - The AI then detects all the price displays present in those shelves. The AI at this stage doesn’t comprehend the meaning of the price display on the shelf. 
  • Step 4 -  The detected price display is fed into the Optical Character Recognition (OCR) engine to comprehend the meaning of it.
  • Step 5 -  Then comes the function of the AI layer that finds which product is close to which price display and then attributes that price to that product.
  • Step 6 - Price display detection is now complete.

KEY THINGS TO CONSIDER FOR SUCCESSFUL PRICE DETECTION: 

There are certain good practices that are associated with each process. Adopting them helps in judicious utilisation of the resource at hand. It means the resources are being used to their full potential and the brand is deriving maximum benefits from it. 

Image Recognition AI systems for price detection also follows this norm. There are certain sets of practices which, when followed , helps the brand to easily benefit from the technology. Some such best practices are :

1. GOOD PICTURE QUALITY: 

A good quality picture is important. Pictures that do not showcase the SKU properly will be rejected by the AI instantaneously. 

What is the meaning of a ‘bad quality’ and ‘good quality’ image ?

Bad quality images are essentially those pictures that are blurred, or are too dim or have glare on them, and they may lack a proper orientation. This makes them difficult to compute.

A good quality picture is the one that is blur free, is neither too dim nor too bright and has the right orientation.

This helps in  properly discerning the SKU captured on the image. Taking good pictures (around 10 megapixels + ) helps to read price displays efficiently. What follows is a well-trained AI consequently leading to more accuracy.

2. CREATING AN SKU PRICE LIST:

It's a good practice to create a database of prices of the SKUs involved. As discussed above, due to, say; a momentary lack in planogram compliance the product shifts its position creating an ambiguous situation in price display allotment. When supplied with a repository of SKU price displays, the AI can dip into this resource and check the estimated price of the SKU against what it has detected.

For example, if confused whether the price is $2.99, $29.90 or $299.00; then knowing that the desired price is $3.00, will help the AI . This training will help AI function better and quicker , delivering near pin-point accuracy over time.

3. ANALYZING PRICE DISPLAYS FOR KEY SKUs AND PROMOTIONS:

As one starts off, it's a good practice to first focus on your hero SKUs and special promotions and then move onto other SKUs of the brand. As the benefits reaped from price detection is calculated, this can be extended to other SKUs gradually, as well.

Having a price display detection in your retail image recognition solution helps the brand to achieve real-time actionable insights. The sales reps’ can then redress undesirable situations that arise from it. Over time, price display detection along with other KPIs helps in creation of robust planograms. Consequently, a positive customer interaction with the brand is observed leading to increased sales and value addition to your brand.

Want to know more about other shelf KPIs? Read our next blog to find out.

To see how your own brand is performing on the shelves, click here to schedule a free demo of ShelfWatch.

Price detection and its compliance is an important part when it comes to achieving a perfect store programme. Brands are looking towards retail image recognition solutions for the same. Monitoring price compliance through price detection is an important feature of our AI retail image recognition solution. In this blog, we discuss why price detection is important and provide a bird’s eye view of how the technology works.

WHY SHOULD CPG COMPANIES MONITOR DISPLAY PRICES? 

The importance of price display monitoring for Consumer Packaged Goods (CPG) companies mainly comes from  high instances where incorrect prices are displayed to the user than what was intended. Some such instances that lead to it are - 

  1. When retailers don’t follow the guided price range.
  2. A missing price display.
  3. An incorrect location of price display.
  4. Promotions (like discounts and combo pack prices) may not be reflected in the price display.
  5. Price display does not reflect the changed pricing.

1. WHY DO INCORRECT PRICES GET DISPLAYED?

The reasons for the above situation could be myriad.

Retailers may cut or raise prices Another reason could be that the retailer might not have an updated database for changed pricing.

Not only this, when the customers move around the store, they may pick up a product and place it somewhere else. This disturbs product placement and consequently, the price display that has been allotted to that product.

Often retail representatives are responsible for handling a lot of products. They have to put in order several products and keep it in sync with the Point of Sale Material (POSM). It's a big task to undertake and could sometimes lead to erroneous execution. All this could potentially lead to incorrectly displayed prices.


2. THE IMPACT OF INCORRECTLY DISPLAYED PRICES

Some retailers may aggressively slash or raise prices. If the prices are raised then there is a loss of sales. If prices are slashed then the company loses revenue. Either of these scenarios do not align with the company's strategy.

Also, a not updated database could lead to inconsistent pricing across different outlets. This would be at odds with your brand strategy of giving a uniform customer experience. This unplanned inconsistency in pricing could also hurt your retailer relationships.

Then there’s the instance of an incorrect price display. For instance, let's say a customer enters a store to buy a shampoo belonging to Brand ‘A’. But when they reach the aisle, they encounter a wrong price display which has been attributed to it. The unsuspecting customer may just assume it's the price of their shampoo and decide to buy its cheaper alternative, say one that belongs to a brand called ‘B’. 

After a few days, this customer requires a conditioner. It makes more sense for them to purchase the conditioner of Brand B which compliments their shampoo. Not to mention that people form a routine and stick to it, which means they are now friends with Brand B.

This causes a direct impact on sales and the strategy that the brand intends to follow in the market. It is also a cause of poor customer experience, This could make sales loss permanent and may also extend to other categories of the brand.

In the same manner, an incorrectly displayed price would negatively impact sales when promotions are being offered by the brand. During promotions, rules for POSM have to be complied with, especially correct price display. A wrong price display here would essentially thwart the entire reason for the promotion.

DIFFERENT WAYS OF MONITORING PRICE:

Different methods of monitoring price tag detection by sales reps, third party audits, retailers and use of AI image recognition
                     

Key Performance Indicators (KPIs) like Planogram compliance, POSM compliance are aided, and their standards improved with the help of price detection feature. Also, CPG companies keep in mind foremost, the issues that a wrong price display can lead to. As a result, they conduct price monitoring exercises as a part of their routine store visits. These visits can be carried out by:

1. Third party Audit: Herein the FMCG in order to implement their perfect stores, creates a set of standards with the help of KPIs. Then they hire an independent auditing company to visit their stores and see if those standards are being implemented.

2. Self reporting by field reps: In this case, the company uses its own field reps who routinely visit stores and they see to it that compliance is established. Wherever it is not, the measured KPIs give them direction to proceed and try to create a perfect store condition for their brand.

3. Retail Partners : Often retail partners are encouraged to gather information regarding price detection data. The reports they generate are used by the brands.

These methods are mostly manual. And us humans? We are subject to bias. 

Self reporting by company sales rep’s is essentially a conflict of interest. They may not relay the ground reality to ensure their sales target is met for the month. 

And as for retailers, especially in general trade, there’s an absence of a standardised system that efficiently catalogues brand sales’ in different categories. Either way, their perspective when storing information is the store’s sales and not brand health.

This has led to the advent of a new player : AI powered image recognition and object detection solution that strives for retail compliance in an objective and measured manner. In fact, as per Gartner report, Image recognition technology can increase sales force productivity, improve shelf condition insights and help drive incremental sales. 

IMAGE RECOGNITION FOR PRICE DETECTION:

This solution involves a software that uses image recognition to identify the brand and category and object detection technology for identifying SKUs. 

Herein , the sales reps or the third party auditors use cameras or mobiles to capture pictures of the shelves. Then these pictures are sent to the cloud server where the AI processes it. It detects the SKU and then calculates the KPIs associated with it. They may include out of stock, share of shelf and price detection.

Understanding the technology in more detail:

Understanding the technology behind price detection by AI solution for retail image recognition

The AI is trained to recognise the brand’s SKUs with the help of images sourced from the FMCG. When deployed on field, it follows the following process in order to perform price detection:

  • Step 1 - The AI firstly detects all SKUs present in the picture
  • Step 2 - The AI then detects the shelves present
  • Step 3 - The AI then detects all the price displays present in those shelves. The AI at this stage doesn’t comprehend the meaning of the price display on the shelf. 
  • Step 4 -  The detected price display is fed into the Optical Character Recognition (OCR) engine to comprehend the meaning of it.
  • Step 5 -  Then comes the function of the AI layer that finds which product is close to which price display and then attributes that price to that product.
  • Step 6 - Price display detection is now complete.

KEY THINGS TO CONSIDER FOR SUCCESSFUL PRICE DETECTION: 

There are certain good practices that are associated with each process. Adopting them helps in judicious utilisation of the resource at hand. It means the resources are being used to their full potential and the brand is deriving maximum benefits from it. 

Image Recognition AI systems for price detection also follows this norm. There are certain sets of practices which, when followed , helps the brand to easily benefit from the technology. Some such best practices are :

1. GOOD PICTURE QUALITY: 

A good quality picture is important. Pictures that do not showcase the SKU properly will be rejected by the AI instantaneously. 

What is the meaning of a ‘bad quality’ and ‘good quality’ image ?

Bad quality images are essentially those pictures that are blurred, or are too dim or have glare on them, and they may lack a proper orientation. This makes them difficult to compute.

A good quality picture is the one that is blur free, is neither too dim nor too bright and has the right orientation.

This helps in  properly discerning the SKU captured on the image. Taking good pictures (around 10 megapixels + ) helps to read price displays efficiently. What follows is a well-trained AI consequently leading to more accuracy.

2. CREATING AN SKU PRICE LIST:

It's a good practice to create a database of prices of the SKUs involved. As discussed above, due to, say; a momentary lack in planogram compliance the product shifts its position creating an ambiguous situation in price display allotment. When supplied with a repository of SKU price displays, the AI can dip into this resource and check the estimated price of the SKU against what it has detected.

For example, if confused whether the price is $2.99, $29.90 or $299.00; then knowing that the desired price is $3.00, will help the AI . This training will help AI function better and quicker , delivering near pin-point accuracy over time.

3. ANALYZING PRICE DISPLAYS FOR KEY SKUs AND PROMOTIONS:

As one starts off, it's a good practice to first focus on your hero SKUs and special promotions and then move onto other SKUs of the brand. As the benefits reaped from price detection is calculated, this can be extended to other SKUs gradually, as well.

Having a price display detection in your retail image recognition solution helps the brand to achieve real-time actionable insights. The sales reps’ can then redress undesirable situations that arise from it. Over time, price display detection along with other KPIs helps in creation of robust planograms. Consequently, a positive customer interaction with the brand is observed leading to increased sales and value addition to your brand.

Want to know more about other shelf KPIs? Read our next blog to find out.

To see how your own brand is performing on the shelves, click here to schedule a free demo of ShelfWatch.

No items found.
Ankit Singh
Co-Founder, CTO ParallelDots
Ankit has over seven years of entrepreneurial experience spanning multiple roles across software development and product management with AI at its core. He is currently the co-founder and CTO of ParallelDots. At ParallelDots, he is heading the product and engineering teams to build enterprise grade solutions that is deployed across several Fortune 100 customers.
A graduate from IIT Kharagpur, Ankit worked for Rio Tinto in Australia before moving back to India to start ParallelDots.