1.      Audience

Audience is at the core of your presentation. You need to understand your audience, their skill-set and their focus area before you start working on your presentation. If you audience is Chief Technology Office or Data Scientists, you may talk about technical jargon and machine learning models; however, if your audience is Finance or Marketing professionals, then talking about Random Forest or Gradient Boosting will be complete alien to them.

2.      Facts

Stick to facts to the extent possible. In the case where you are presenting you opinion, that should be backed by facts and logic. A lot of times, people tend to bring in their bias and judgement based on their experiences. The whole idea behind using data is to remove bias and get a true sense of business. Don’t get carried away by emotions, judgments and experiences – always support your conclusion and opinion with facts.

3.      Structure

Create a skeleton of how and in what manner are you going to create your presentation. When I have to create a presentation, I always start with a clean slate and start adding titles to ‘Table of Contents’ or create a slide and write what you are going to cover in each of those slides. Once you have created a skeleton and added high-level points for each of the slides, you can go through each of the slides and prune to make the final structure. This will help you in achieving three objectives:

1. This will limit your tendency to provide redundant information.
2. Your information flow will be more coherent and organized .
3. Likelihood of missing out any key point will be reduced.

You need to ensure that data points and analysis are coherent with each other; one point should lead to another – in data analysis, one analysis should lead to another analysis.

4.      Exhaustive

Collectively, your presentation should be complete in itself – anyone going through your presentation for the first time should get clear understanding of the project, background, approach, results and impact, along with proper reasoning and explanation. This was covered in detail in the first part of this series: Why Storytelling is One of the Most Important Things for a Data Scientist?

5.      Action Title and Call-outs

Action title is one of most important components of the slides that I make. If I were to explain the entire slide or key takeaway from the slide in a single line, it should be reflected in the action title. In the scenarios where slides are too loaded with text or numbers, senior level executives don’t prefer to go through each of the points; instead, they are more interested in key takeaway from the slide. If you can’t express the entire slide in one line, you need to work on that slide, either it has too much information or too less of an information. Call-outs should be used to highlight any kind of outlier or abnormal behavior. You may just highlight/border the text with a different color and provide explanation for the same. In cases where you want to explain some behavior/trend in a graph, create a call-out and write your explanation for the behavior. It comes very handy while we are doing diagnostic analytics.

6.      Frame of Reference

Provide a frame of reference while emphasizing your analysis or results. For example, if I were to say, “sales of my company this year were US 110 Mn”, this standalone statement may not give me complete information. A better statement would be “sales of my company this were US 100 Mn, which represented YoY growth of 15%.” Second statement creates more impact in the mind of audience. Let’s take another example – “because of improved accuracy, revenue impact of the model will be US 10 Mn” vs “because of improved accuracy, revenue impact of the model will be US 10 Mn, which is 5% of our FY2018 revenue.” Bringing out these frames of references at the right places in your presentation make them more impactful.

7.      Icons

As it is said that visuals are always better than text, so why not use that to our advantage. Microsoft Office offers beautiful and rich collection of Icons that we can use to highlight certain points. For example, if I were to explain the scale of project that I worked upon, below are two methods I can use:

Method 1:

“The complexity of the project can be gauged from the fact that we had to deal with 1,000 Trucks, 100 million data records, 100,000 trips per year.”

Method 2:

Which method would you choose?

8.      Smart Art

MS Office provides some wonderful smart arts which you can use to present your points in a visually appealing manner. For example, if you want to present the entire project flow or different stages of your engagement, you may use process flow smart art. You may create your own smart arts using icons and shapes. In another example, let’s say you want to present a process which has data consolidation, followed by analytics layer and ending with a visualization layer – you can use a combination of icons, shapes and smart arts.

9.      Color Scheme

Be very careful of the color scheme that you use. Most of the companies would have their defined color scheme; in such cases you should adhere to the prescribed scheme. In case there is no prescribed color scheme, make sure that you don’t use very light or very bright color. Keep a consistent theme with your colors; if you want to highlight certain points, use same color across your presentation. For instance, you may use red color to highlight negative points or green color for positive points.

10.  Independent Feedback

This is an additional step which comes in when you have done everything at your end. Basically, once you have completed your presentation you should get it reviewed by a third person and take an independent view (this may not be possible in cases where you are working on highly confidential projects). Independent views will give you a clarity if you have covered all the aspects in terms of the presentation being exhaustive and coherent. Ensure that you are open to receiving feedback and inputs and justify them with objectivity, wherever required.

Learn everything about RPA for FREE in this course – Introduction to Robotic Process Automation (RPA)

The post Presentation for Data Science Project – 10 Points to Master appeared first on AI MONKS.

In my last five years of working in analytics and consulting, I have come across this phrase thousands of times:

“The story is not coming out properly.

We need to work on story first, rest can be done easily.”

Why so much emphasis on story? After all, in data science and analytics what matters most is the accuracy of the model, isn’t it?

Are you sure?

Let’s figure that out.

Let’s assume that you have worked on a churn prediction model in your organization and you have to present the entire engagement to your CEO and CFO, who don’t have any background of the project. Seems easy? Since it’s a data science project, all you have to do is tell them about the model and the accuracy, that’s it. Is it?

Let’s take different scenarios of presenting our results and see which one you will prefer.

Approach 1:

“We have developed a churn prediction model with an accuracy of 82.7%.”

Approach 2:

“We developed an ensemble of random forest and logistic regression to predict the likelihood of a customer churning out in next three months. We have achieved an accuracy of 82.7% using the ensemble. The previous model that was based only on logistic regression had an accuracy of 71.1%.”

Approach 3:

“Our North India zone was facing the challenge of high churn rates. A lot of old customers were churning out from our network. We undertook this engagement with the objective to predict customers who are likely to churn in next 3 months. The existing model lacked the accuracy that we wanted to achieve. For this, we used internal customer data and complemented it with external social media and demographic data to develop an ensemble of model. We developed a random forest model with 1000 trees and logistic regression which is based on maximum likelihood estimation. This model gave us an accuracy of 82.7% which is better than that of our previous model.”

Approach 4:

“Our North India zone was facing the challenge of high churn rates. For the past four quarters, churn rates were over 10% each quarter which was leading to a revenue loss of around USD 100,000 per quarter. A lot of old customers were churning out from our network.

We undertook this engagement with the objective to predict customers who are likely to churn in next 3 months. By estimating this probability, we would want to focus on customers (through offers, promotions or improving customer relationship) who have high likelihood of churning out.

Though we already had a model for the same, but the accuracy provided by the model was not helping us much. So, we developed a new algorithm that combines two machine learning models and provides us better accuracy (82.7%) than the previous model (71.1%). For developing this new model, we used internal customer data and enriched it with external social media and demographic data to develop an ensemble of model. The overall impact of this model is estimated to be around USD 75,000 per quarter.”

Which approach would you choose if you were to explain the engagement to your CEO and CFO?

Approach 1? Hell, No…

Approach 2? No…

Approach 3? Umm… May be

Approach 4? Yes, certainly.

Now, one can always argue that approach 4 is nothing but providing complete details about the project – background, data, algorithm, output and impact.

That’s what storytelling is.

Five Key Aspects of Storytelling

One of the most important aspects you need to consider while providing all these details is your ‘audience.’ Understand who your audience is. If your audience is Chief Data Scientist then you should go into technicalities of the model; however, in this case our audience is CEO and CFO. They would be least interested in the underlying model, rather, they are more concerned about the revenue/cost impact it will have on business.

I will define storytelling as “presenting different aspects of a project in such a manner that they are exhaustive, coherent, succinct and audience-friendly.”

Now, let’s understand different aspects of a data science projects and learn how to present them in your story.

Any data science or data analysis project would essentially cover five different aspects.

Can you relate this with the ‘Approach 4’ mentioned above? Is there any other Approach which covers all the five aspects?

No, right?

When you are working on large datasets and your aim is to develop a model that improve accuracy, it is not uncommon to lose sight of the larger picture. That’s why while presenting your results, you should keep the above framework in your mind and build your entire story accordingly.

Your model may be among the best models in the world but if you can’t convince business users about it, they will always have their apprehensions in implementing that. So, you should give due attention to first and last aspect of the above framework – ‘Background’ and ‘Impact’. These two aspects will help business users understand the criticality and importance of your project.

If you like this post, we are sure that you will also like – Presentation for Data Science Project – 10 Points to Master

The post Storytelling for Data Scientists – Why is it important? appeared first on AI MONKS.

This post is a continuation of the first post Sales & Marketing Analytics – Top Analytics Projects in Retail Industry. In the first post, we talked about the function which is at the forefront of every business and has a direct impact on the top line of the company. This post talks about the other functions of the business which may not be at the forefront but forms the backbone of the business. Operations and supply chain are as essential functions, if not more, to the business as the sales & marketing. Sales & marketing focus on selling products to consumers; while, operations and supply chain focus on producing those products and ensuring that the products are accessible to consumers.

There are five different sub-functions that we will discuss under operations and supply chain function. Then under each of the sub-functions, we will talk about top analytics use cases to make the business more data-driven.

The five sub-functions in operations and supply chain are as follow (refer to the image):

1. Demand Planning
2. Procurement
3. Inventory
4. Logistics & Transportation
5. Vendor Management

Moving a step further, let’s look at the top supply chain analytics use cases spread across these five sub-functions.

A) Demand Planning

1. Forecasting

B) Procurement

2. Reorder Level Identification

3. Price Variance Analysis

C) Inventory

4. Inventory Rationalization

a) 9-Box Analysis

b) Variability (sigma/mu) Analysis

5. Stock-out Prediction

D) Logistics and Transportation

6. Fleet Monitoring

7. Network Planning

8. Route Optimization

E) Vendor Management

9. Vendor Scoring

10. Vendor Consolidation

Let’s now talk about each of the use cases in detail and understand how and why they are relevant in the business context.

1. Forecasting:

Forecasting is one of the most essential parts of demand planning. Forecasting sales (based on primary or secondary or both datasets) is essential to maintain a balance between meeting consumers’ demands and not get bogged down by excessive inventory. Traditionally, forecasting has been done through rudimentary approaches or based on one’s judgement. However, with advancements in technology and sophisticated algorithm, it is possible to do forecasting at SKU level with decent accuracy.

2. Reorder Level Identification:

Once you have identified (forecasted) the number of goods you need to procure, identifying the right reorder level is essential to ensure that the production doesn’t stop because of stock-outs, as well as ensure that working capital is not blocked because of incorrect orders. Historically, procurement managers have been identifying reorder levels at a product category level based on their judgements; however, now it has become convenient to calculate reorder level for each item that is procured. Moreover, based on historical data one can also identify the demand distribution curve that different products have and orders for each can be placed accordingly using statistical techniques.

3. Price Variance Analysis:

This may not be a very common use case, but this essentially provides procurement teams to identify cost savings opportunities. Price variance analysis means analyzing which all items were procured from which all vendors and at what costs? Are there any significant disparities in unit economics? Are we buying the same item from different vendors at different prices? Is there a business rationale for a higher price? Is this an indication for kickbacks or any kind of frauds by the procurement function? This analysis, though mostly descriptive and statistical in nature, plays an important role in identifying cost savings opportunities and uncover any fraudulent activities that may be taking place.

4. Inventory Rationalization:

This is one of the most common and most useful use cases of analytics for the operations function. Any amount of unsold inventory is blocked working capital for the company. The unsold inventory doesn’t just use the limited amount of space that you may have in your store/warehouse, it also has a significant impact on your working capital. Imagine you are a billion-dollar retail enterprise and you find out that you have an inventory worth USD 25 million lying in your warehouse which has not been sold for over a year now. Do you know how much money are you losing because of that? If you were (or if you already have) to borrow that amount of money from a bank at 6% interest rate per annum, you will be paying USD 1.5 million as interest. Assuming your net average profit margin for products in that USD 25 million inventory is 8%, your total margin would have been USD 2 million. Now, after a year you have already lost USD 1.5 million as interest, you are just left with USD 0.5 million as your profit which is 2% margin. That’s why inventory rationalization is one of the most common and most sought-after use cases in the industry.

5. Stock-out Prediction:

According to Retaildive.com, retailers are loosing around USD 1 trillion annually because of out-of-stocks (Link). Identifying which SKUs are likely to stock-out at what time is the key to handle this problem. This is not a stand-alone use case but has to be carried out with efficient forecasting and right reorder level calculation (Use Case 1 and 2). Avoiding stock-outs does not just help in preventing lost sales, it increases customer loyalty as well. If a customer visits a store and every time she finds what she is looking for, the likelihood of repeat visits increases considerably.

6. Fleet Monitoring:

Fleet monitoring is mainly a diagnostic exercise that companies carry out to understand the movement of their fleet, esp. in the cases where companies have owned or leased fleet. This helps companies in evaluating if their vehicles are under-utilized or over-utilized or adequately utilized. Also, the performance of fleet at a trip level, route level and by a driver can be assessed. Furthermore, the analysis at a trip level can help companies identify performance improvement opportunities in their fleet movement by following best practices.

7. Network Planning:

Network planning is an optimization problem carried out by companies having a presence across multiple-states or large geographical area. The objective of this problem is to identify the right locations of warehouses or distribution centres from where a company can ship goods to its distributors or dealers. In this optimization problem, the objective function is to minimize cost while meeting all the SLAs (service level agreements) signed with dealers. This exercise is usually carried out by companies once a quarter or half-year, depending on the growth trajectory, or when new locations are to be identified.

8. Route Optimization:

Route optimization is again an optimization problem but mainly carried out on run-time for truckers delivering goods to different locations. Route optimization is an exercise carried out to identify which route should a vehicle follow to deliver goods to intended locations. The inputs that go into this problem are the delivery loads (or orders from different locations), truck capacity, the distance of delivery locations from distribution centres and time constraints (defined from SLAs). Basis these inputs, our objective is to define a route which a vehicle should follow such that the cost is minimized, and all the deliveries are made within the stipulated time.

9. Vendor Scoring:

Vendor scoring, though descriptive in nature, is a very useful application of analytics for companies to score their vendors based on multiple parameters such as quality of products and service, adhere to service level agreements, the price charged for products and services, the criticality of a vendor in the overall functioning of the business. This is mainly carried out by companies which have thousands of vendors (manufacturing companies, for instance). This helps companies identify their top vendors objectively and develop strong relations with them to ensure business continuity.

10. Vendor Consolidation:

Vendor consolidation is another cost-saving technique followed by a lot of companies (primarily by companies having thousands of vendors, as in the manufacturing sector). The objective of this exercise is to identify if there are any opportunities to reduce the number of vendors that the company may have and by giving more business to a limited set of vendors, negotiate better prices or terms of business. Terms of business could mean shorter lead times or longer payment period. The latter part will help the company optimize its working capital.

All the projects listed above may not essentially be data science and machine learning projects, but one thing to note here is that the simple descriptive analytics can also provide multiple use-cases for companies which have a significant impact on business operations. Also, with the volume of data that is being generated these days, it is becoming more and more essential to use analytics tools even for slicing and dicing data.

We would love to hear your experiences with any of these projects. If you have done any other project that is not listed here, we would be happy to learn from your experience. Please free to reach out to us at info@aimonks.com.

The post Supply Chain Analytics – Top 10 Analytics Projects appeared first on AI MONKS.

Banking analytics seems to be a promising solution in such a case.

The financial services industry is one of the industries which has embraced analytics across the entire customer lifecycle. Right from identifying products for a customer to designing promotional campaigns, data science has been adopted at each customer touchpoint. From underwriting to delinquency, financial institutions are adopting data science and analytics across different functions. In this post, we have highlighted the top 10 analytics projects used by the financial services industry to attract and retain customers. You may try your hands on any one of these projects.

Lead prioritization:

Lead prioritization is used where ticket value is considerably high and resources to sell are limited. In the BFSI sector, the ticket may refer to a high-value loan (home/auto/personal/corporate/etc.) or a policy or an investment scheme. Accordingly, customers can be segmented into different categories based on their likelihood to buy the product. Lead prioritization helps in allocating limited resources to potential leads with a higher chance of a conversion. The more technical term for this is ‘customer propensity modelling’.

Customer lifetime value (CLV):

Customer lifetime value is a value that the customer is likely to provide to the bank in his entire relationship with the bank. This helps the bank in evaluating the investment and effort that the bank should put in to acquire and retain the customers. Lead prioritization and CLV estimation serve a similar purpose of identifying top leads and customers, respectively.

Predicting the life event of a customer:

Based on a customer’s bank statement or credit/debit card statement, one can predict future life event of a customer. Since the banks already have this data for customers, data procurement will not be a task for the bank. Based on the predicted life event, relevant products can be offered to customers. An important point to note here is that the tagging of transactions will play an important role in determining the accuracy of the model. Additionally, in the case of salary accounts of customers, looking at the salary trend of the customer can provide information about a recent promotion or a job change.

RFM modelling:

This is a customer segmentation technique based on three parameters – recency, frequency and monetary. Recency refers to how recently the customer has purchased a product; frequency refers to the frequency of purchases by the customer; while monetary refers to the money value of purchases by the customer. Customers in each segment would usually possess similar attributes and can be served with similar kind of promotional offers.

Cross-selling/up-selling:

Cross-selling/up-selling refers to selling additional products to existing customers. These products could be different from the ones that the customer has already bought or top-up of existing products. An example of cross-selling for a bank is offering a credit card to a customer having a savings account with the bank; while, an example of up-selling is increasing (topping-up) the insurance cover of the policy-holder. Again, in this case, if you can predict the current and future life event of the customer, the probability of cross-selling the right product increases considerably.

Learn about Sales & Marketing Analytics – Top Analytics Projects in Retail Industry

Next best offer:

Next best offer is used for improving promotional efficiency and increase conversion rates at different customer touch points. Based on a customer’s historical behaviour, the company can send targeted promotions to which the customer is likely to respond. This is another method of cross-selling based on a recommendation engine. Again, having details of customer’s current and expected life events can improve the likelihood of offering the right product.

Sentiment Analysis:

Carrying out a sentiment analysis on social media posts can provide detailed insights to customer feedback. Not just social media, sentiment analysis can be carried out on support emails, app reviews or any other customer touchpoint. Furthermore, this can be extended to carry out a lot of other analysis such as emotion analysis, key negative/positive themes, topic modelling, etc.

Delinquency prediction:

NPAs – Non-performing Assets: this term has become very common in the banking world because of rising defaults in recent years. An account is considered NPA if it defaults on its payment schedule. Default (or delinquency) prediction is one of the most common and useful use cases in the industry. Companies, almost on a daily basis, are trying to improve their models to predict delinquent customers. This helps banks to control the revenue (principal + interest, both) loss against an issued loan.

Do you know restaurant industry is using analytics to gain a competitive advantage? Read Restaurant Analytics – Top Analytics Project Ideas to know details.

Predicting the probability of a customer to renew insurance (or any other) policies:

This is very useful for insurance companies. Since the insurance companies don’t have access to customer’s banking history, achieving accuracy becomes challenging only with internal data points. The model for this can be enriched by adding external data points related to the customer’s current life stage. If policy issuer is the same as the bank where the customer holds his account, the prediction model can be enhanced with customer’s banking data.

Churn Analysis:

The technical term in banking parlance for this is ‘balance transfer.’ Balance transfer refers to transferring of live loan account by a customer from one bank to another, mostly because of better interest rates or improved service. Churn analysis provides knowledge about customers who are likely to transfer the balance to other banks. Subsequently, as a remediation measure, the bank can take corrective measures such as building a strong relationship with customers or offering them other products at discounted prices to prevent customer churn.

BFSI sector provides a plethora of other customer analytics projects that you can take up.

Have you worked on any of these projects or maybe, a different one? We would love to hear your views and approach to carrying out these projects. Don’t forget to share your experience with us at info@aimonks.com.

The post Banking Analytics – Top Analytics Project Ideas appeared first on AI MONKS.

With the rising competition and increasing customer expectations, it has become difficult and costlier to acquire new customers and retain existing customers. That’s where sales & marketing analytics has provided a ray of hope to marketing & sales teams in organizations. In this post, we will talk about some of the high impact sales & marketing analytics projects which can be quite useful for marketing & sales functions of organizations in retail industry. Other industries can also leverage these use cases with little modification.

1. Recommendation System:

This use case refers to developing a recommendation system to recommend products (or movies or songs) to customers. This is similar to the recommendation system we see on Quora, LinkedIn, Netflix, Facebook, Amazon, Flipkart or any other e-commerce website. There are two types of recommendation systems: content-based and user-based recommendation systems. This use case is very prominent for companies operating in the e-commerce or online space. This helps companies in increasing share of wallet of the customer by engaging them with recommended products.

2. Lead Prioritization: 

Lead prioritization is useful for companies during the sale of high-value items such as cars, flats, etc. You can prioritize customers based on their likelihood (propensity) to buy the product. The more technical term for this is customer propensity modelling. Companies in real estate use this technique to prioritize customers based on their likelihood to purchase a property. Also, automotive companies have developed customer propensity models for their dealerships wherein a sales agent inputs certain data points and predict the likelihood of converting a lead into a customer.

3. Market Basket Analysis: 

Market basket analysis is used for two major purposes. Firstly, it is used to improve share of wallet from existing customers by cross-selling and up-selling. Secondly, it is used to define the placement of products in a brick-and-mortar store. Though this has its own importance for brick-and-mortar stores, companies in the e-commerce space are using this technique for bundling of products. Products which are frequently bought together are bundled together and sold at a certain discount or with some offer to increase average transaction value.

4. Next Best Offer: 

This is similar to the recommendation system but more from the perspective of rolling-out promotions and offers to customers. This technique is used to improve promotional efficiency and increase conversion rates from promotions. Based on a customer’s transaction history and response to past promotions, the company sends targeted promotions which a customer is likely to respond, eventually resulting in a monetary transaction. This is used by both e-commerce companies as well as brick-and-mortar stores. It is mainly based on rolling-out offers on digital media such as emails, Facebook, Instagram, etc.

5. Channel Attribution Modelling: 

This is particularly useful for an e-commerce company to identify which online channels are driving traffic to the site. Since promotions are carried out on multiple digital channels, it becomes imperative for companies to identify which all channels are driving traffic. Based on this knowledge, companies can redefine their advertising strategy and look at reallocating advertising budget to the channels where most of the [potential] customers are present.

You may also like Restaurant Analytics – Top Analytics Project Ideas

6. RFM Modelling: 

RFM is essentially a segmentation technique based on recency, frequency, monetary. Recency refers to how recently the customer has purchased a product; frequency refers to the frequency of purchases by the customer; while monetary refers to the monetary value of purchases by the customer.
The idea here is that customers in a segment tend to have similar buying patterns and attributes. So, companies can target customers in a segment in the same fashion. For instance, one segment can be defined for customers who buy products worth INR 10,000-12,000 every month. The second segment can be a set of customers who used to buy items worth INR 10,000 every month but haven’t bought anything in the last 3-4 months. Hence, one type of offer can be rolled out to one segment of customers; while a different type of offer can be sent to the customers in second segment.

7. Sentiment Analysis:

This use case is relevant for any and every company which has a presence on any kind of digital media. Social media is no more a platform merely to communicate with your friends, it has evolved into a stronger platform where customers give their feedback and talk openly about their experiences with a brand. Considering the sheer number of people on the internet (translating to the number of customers), it becomes inevitable to track the sentiments of customers and convert them to promoters.

8. Marketing Mix Modeling:

Marketing mix modelling is used to optimize the marketing spend in order to have a maximum positive impact on sales. It essentially includes mapping the impact or effectiveness of various marketing initiatives to the sales of the company. This can include any kind of channel, online and offline, where there was a marketing spend.

9. Customer Lifetime Value (CLV) Analysis:

CLV is a monetary value that a company expects to earn from a customer during the entire customer journey. This analysis has become important in today’s world because companies’ marketing and advertising budgets are shrinking. In such a scenario, the company needs to be prudent in focusing energy only on those customers who are likely to return high value back to the company. For instance, the cost of acquiring a customer is INR 1,000. There are two potential leads: from the first lead, you expect to earn INR 1,500 in the next 3 years; while, from the second lead you expect to earn INR 10,000. Whom would you focus your energy on?

10. Loss of Sales Analysis:

Though this is a very important analysis, very few companies work on lost sales because of limited data availability. Loss of sales analysis refers to analyzing lost sales due to varied reasons. For instance, stock-outs or unavailability of required size in case of apparel could be one reason; while high waiting time in the billing queue could be another reason. High waiting time in the billing queue may result in poor customer experience; thereby, leading to customers not buying the product at all. There could be multiple other reasons for lost sales. This analysis can help companies identify the improvement areas for companies to reduce lost sales. Furthermore, insights from this analysis can have a positive impact on the company’s top line and improve customer experience.

Retail companies carry out multiple other sales & marketing analytics projects for acquiring and retaining customers. Have you done any of these projects? Or, maybe a different one?

We would love to hear your experiences of doing customer analytics projects. Please share your experience us at info@aimonks.com.

Don’t forget to read out the second part: Supply Chain Analytics – Top 10 Analytics Projects

The post Sales & Marketing Analytics – Top Analytics Projects in Retail Industry appeared first on AI MONKS.

Did you know this insight?

Imagine if you have this and tens of other such insights that have a direct impact on your business. This is the power of restaurant analytics and the impact it can have on your business. The commoditization of hardware and software has made capture and analysis of data easy and economical. Also, due to increasing competition and rising customer expectations, it has become increasingly important to capture data and personalize the customer experience and use data for business performance improvement.

The restaurant industry is one such industry where competition has increased and at the same time, customers have become even more demanding. With the onset of cloud kitchens, increasing investment from private equity and emergence of food delivery apps, the competition in the space has become fierce for all sorts of players in the market, be it full-service restaurant, quick service restaurant or a well-recognized chain. In such a scenario, data analytics can play an important role in creating a differentiating factor for a restaurant. Better knowledge of business, personalized customer experience, reduced wastage and pilferage and targeted marketing are some of the key areas where data analytics can help in either increasing revenue opportunities or reduce the expenses.

We have listed down 8 top data analytics use cases that are essential for any restaurant business for data-driven decision making.

1. Forecasting for Restaurant Business:

Forecasting is an essential part of any business. Based on the sales forecast and customer inflow, things need to be planned to provide a smooth experience to customers. In the restaurant business, forecasting can be done at three levels:

A) Sales Value:

Forecasting how much are you going to earn in the next three days or next week or next month is essential to plan your budgeting activities and manage your P/L effectively. Usually, this is high-level forecasting for the business as a whole. There are multiple forecasting techniques one can use, the simplest of them being linear regression. The more complex are non-linear regression models, neural networks and others.

B) Customer Visits:

This is again essential to understand when (day of the week and time of the week) your customers prefer to visit. You need to understand what day of the week and what time of the day is preferred by your customers. You can optimize your staffing requirements and inventory accordingly. In the restaurant industry, this would mean forecasting table turns by day of the week and time of the day.

Below chart gives an idea about the visit of customers – split by day of the week and time of the day.

The above analysis provides us with insights into the visit patterns of customers. What day and what time is preferred by customers – combining this data with order details and customer details [Refer Use Case 8 – Customer Segmentation in this post], you can go a level deeper and identify which segment of customer prefer kind of food and at what time they prefer to visit.

C) Delivery Orders:

With the emergence of food delivery apps, most of the restaurants are trying to improve their top line by encashing on the orders placed by customers through these food delivery apps. Predicting how many and what items will be ordered in any given hour will help in reducing the overall order time, thereby improving customer experience and hence, customer loyalty.

We have discussed different kinds of forecasting which are useful for the restaurant business. One important point to note here is that there are too many external factors that could impact the accuracy of your forecasting model. Some of the external factors that could have a significant impact on your restaurant business are the weather, office holidays, school/college holidays, exams, weekday/weekend, among others. Due care should be taken to include all such factors in your forecasting model.

2. Menu Engineering:

Menu engineering refers to creating the right mix of menu keeping into consideration the customers’ choices and preferences, and profit margins. Based on order history, customer feedback, pricing and other factors, the restaurant may identify the food items ordered and liked by customers. Menu engineering can be done to the level where you identify which items are sold in breakfast, which are sold in lunch or dinner. The menu can be altered in a manner that the items which are not sold or sold rarely during dinner time are removed. This will help in reducing the wastage of food at the end of the day, thereby, reducing cost. Below is a chart which shows a similar kind of analysis.

Key insights that can be derived from the above analysis:

There are tons of other useful insights which could be gathered from this chart. That’s what this analysis one of the most important analysis for any restaurant business.

3. Customer Feedback Analysis:

Analyzing customer feedback in the restaurant business is as essential as anything else in the entire operations. Customers tend to post photos and reviews on Instagram, Facebook and restaurant-aggregator websites. These reviews need to be properly mined and analyzed for customer’s preferences, feedback, likings and dislikings. You may have an account on Facebook, Instagram, Twitter and others but are you reaching out to right customers? Are you listening to your customers? There are multiple tools available in the market which can help you carry out detailed analysis on social media data.

4. Location Identification for New Store Opening:

Rising real estate costs combined with mushrooming of different restaurants have made it difficult to identify and afford the prime locations for your stores [Real estate cost for a restaurant business is between 5-10% of gross revenue]. In such a situation, it becomes essential to identify alternate locations where your target customer segment is present, as well as rental expenses, are not as high as prime locations. For example, if your restaurant targets office goers who have their lunch outside on a daily basis, then you should identify locations where new offices are opening up or new business parks are being set up. Initially, the population in that area may not as high but the benefits of the low rental cost and low competition will take care of it.

5. Sales Cannibalization Analysis:

If I open a new store within 2 miles of an existing store, will it impact the sales of my existing store? If yes, how much? Answering these two questions is the crux of sales cannibalization. It’s not just analyzing the impact of new stores, but also considering the impact of new product launches on existing products. There are no specific models to carry out cannibalization analysis, but you can use statistical approaches (such as t-test) to compare the KPIs (such as sales, average value per order, average number of orders per day, etc.) to see if there is statistically significant difference before the new store and after opening of new store.

6. Staff Optimization:

Staff cost for a restaurant business is around 20-30% of gross revenue – that’s huge. Almost one-third to one-fourth of your gross revenue goes into labour cost. Imagine if we were to optimize this cost and lower it by 2-3% with better planning (and not impacting customer service), it has a direct impact on your EBITDA (bottom-line). There are multiple considerations into optimizing staff, most important being full-time vs contractual workforce. This is more of a business decision and depends on the kind of restaurant service you are into – fine dining vs quick service vs café. No matter what it is, you need to play to both kind of workforce and maintain a balance that your cost is controlled, and customer service is not impacted.

7. Bundling of Items:

You must have seen McDonald’s offering a combo pack or Happy Meal – that’s nothing but the bundling of items based on their likelihood of being sold together. The idea here is to sell a combination of items at a discounted price as compared to the sum of individual prices of the items. It may look like we are losing out on revenue by selling on discount, but in reality, it’s increasing the share of wallet of customers. For example, if we were to have two items – burger (costing $ 1.5) and soft drink (costing $ 1) – and if we create a combo for the two and sell it as $ 2 then the chances of people buying both will be way higher. Though this may slightly reduce gross margins, the overall impact due to increased revenue takes care of margin reduction.

You may also like Retail Analytics – Top Analytics and Data Science Project Ideas

8. Customer Segmentation Based on Restaurant Visits:

Understanding your customers is the key to any business; the restaurant business is no different. In fact, it is even more important to understand customers, their demographics, their liking/disliking, choices, preferences, what time they prefer to visit, and further details. Think of yourself as a customer and you visit a restaurant where you have already visited in the past. The restaurant offers you a complimentary cheese pizza because they know you love pizza [How they know you love Pizza and not something else? They have captured your order history where you have ordered pizza 5 times in your last 6 visits.]. Wouldn’t that make you happy? Wouldn’t that increase your loyalty to that restaurant? Let’s understand the customer visit pattern to a restaurant by gender.

A) Day-of-the-week and Time-of-the-day Analysis

The above chart shows a pattern of customer visits to the restaurant. This does not split by gender or any other customer attribute. We can see that we have a high number of customers visiting the restaurant during certain time periods on certain days. Let’s break the same analysis by gender and see if there’s any difference in the trend for men and women.

In the above two charts, we can see the difference in the visit pattern between men and women. Key insights and actions that should be taken based on those insights are as below:

• Men and women, both tend to visit on Sunday afternoon. Promote family/couple lunches on Sunday by offering a customized menu or discounts on table reservations.
• Women prefer to visit on weekdays (Tuesday/Wednesday/Thursday) during lunch hours (1:00 PM to 4:00 PM); while the visits by men during that time is not high as compared to their visits during other times. Check if any promotion was rolled-out only to women customers which resulted in an increase in women customers during this period. If yes, promote such offers during other time periods as well. If not, reach out to women with similar demographics.
• Men prefer to visit on Friday evenings, possibly after office-hours. Look for order details during this time. Are they majorly drinks? Can food items be combined with their drinks order? Can there be certain promotions for couple entries?

Do you know anyone who runs a restaurant business? You can reach out to them and ask them to provide you with data in return for a wonderful analysis. You will get a good hands-on practice in return.

Get access to AI Monks Master Course: Applications for Analytics in Different Industries for FREE.

The post Restaurant Analytics – Top Analytics Project Ideas appeared first on AI MONKS.