Data-driven insight. Insight-driven outcomes. Restaurant Analytics: Proof of Concept Case Study Client Overview Industry Challenges Client Background Colombo is flooded with a lot of restaurants, be it fine-dining, usual dining and take-out joints, fast food restaurants or food trucks. This boom in the industry has created a lot of challenges for the players in the industry. Our client has been in operation from the early 1940s, and have been regarded as a leading restaurant. Coming in from the long-term success of its operations, it has now opened four new branches. Financial When opening a new restaurant or expanding, large sums of capital is needed to cover leasing a building, hiring a new wait and hostess staff, stocking the kitchen and bar areas and buying furniture and decorations. Most restaurants do not turn a good profit for several months or years after opening. Planning To be successful, a restaurant needs a good location, a theme or style that appeals to a broad range of customers and a solid menu. Restaurants must stand out from the crowd, especially restaurants in urban areas with high competition. Even established restaurant owners must continually review their menu and theme to ensure they are giving customers what they want. Managerial Many restaurants have a high staff turnover rate, which can lead to scheduling problems and a stressful work environment. Due to the flurry of new restaurants springing up within Colombo, the restaurant outlets are having heavy competition. Our client had done minimal changes in the past to its structure of menu items and this was an acid test to revamp / rejuvenate and optimize their menu with the customer in mind. The restaurants have implemented sophisticated data capturing system in place and wants to capitalize on collected information to improve decision making processes which will help the client plan further expansion, and also to re-engineer its processes with improving performance as the main objective (main goals: increase revenues, lean/cost-efficient operation, improve customer engagement). This project was run for a period of six weeks, with an extensive analysis run on our client’s data from their four existing branches at that point of time. The client is currently engaged with Argyle X on follow-up projects Competition A restaurant must know how to market its business, how to bring in new customers and how to develop a repeat clientele. 2 Actions Need to engage customer better Actions Insight Few categories drive revenue Consider possibility of having certain outlets as Delivery/Take-Away only Assess network and coverage Actions Insight Delivery is a key revenue driver Need to examine footfall around location at specific times – e.g. if an outlet is in the central business district, what is the potential footfall at lunchtime? How can we better attract these customers Is parking facility an issue? Lunch Time Dine-In Promos – e.g. at certain outlets Take-Away is popular at lunch time. Encourage more Dine-In with special offers on specific days (as evidenced in the data). Credit Card offers for Dine-In – e.g. Visa accounts for 15% Dine-In payments. Promote slow moving items via offers/consider removing or modifying these items Menu Optimization (for further details see Slide 8) Actions Insight Dine-In has highest average ticket size but is lagging at certain outlets Insight Project Summary Loyalty scheme (for further details see Slide 5 and Slide 8) Social Media Customer Satisfaction Assessment – e.g. via surveys 3 Project Overview Increasing customer engagement within better customer profiling Comprehensive assessment of restaurant using internal data and statistical analysis to improve sales Determine KPIs of success Data Sources Data Gathering & Baseline Dataset Creation Modeling Customer Purchase Behavior DL channel main revenue driver Data Extraction Revenue Metrics • • • • By Labor Hour By Seat Hour By Square Footage Average Ticket Size Outlet + Location Analysis Channel Assessment • Delivery (DL) • Take Away (TA) • Dine-In (DI) Customer data What data resides in what system? Transaction Data Data Audit & Cleaning Inventory and Costing data Menu Item data Call Center Data • Does the data make sense? • Is the data clean and match across systems Data Consolidation Consolidating data from all data sources Insights & Recommendations Model Drivers of Revenue Detailed Descriptive Statistics DL/TA only outlets? DI largest average ticket size Outlet-based strategies to promote DI at specific times Few categories drive revenue Cluster Analysis Menu Optimization Gap Analysis Increase Engagement • Loyalty scheme • Social Media • Survey campaign 4 Selected Results – Part I Gap Analysis: Benchmarking against Industry Standard Practices Our analysis was limited to data provided for a period of 1 Month (18th May – 17th June). As a result, weAreas are unable a holistic picture of the performance of However, scale, we recommend which to theprovide restaurant is lagging behind – benchmarked onCDC. a comparable i.e. CDC implement these next steps,number so that of weoutlets could provide a complete assessment of the CDC restaurant chains with similar and reported revenues. provide more comprehensive solutions in the next phase of our engagement. 3 2 1 Social Social Media Media Analysis Analysis Provides Provides insight insight into into customer customer preference preference and and gives gives the the ability ability to to benchmark benchmark performance performance against against competitors. competitors. Potential Potential to to run run promotions promotions Customer Survey – Assess the gapthe between Assess gap between experience and expectation. Possibility to find reason behind low dine-ins and gain insight into customer preference. Loyalty CardCard useful in Useful in findingcustomer specific finding specific customer information. information. Enables the Enables the tailoring of and tailoring of promotions promotions and an inan in-depth understanding depth understanding of of customer behaviour. customer behaviour. 5 Selected Results – Part II Descriptive Statistics Revenue Share by Outlet All Revenues Orders Share by Outlet Outlet 1 All Orders Outlet 1 Outlet 2 Outlet 2 Outlet 3 Dine In Revenues Outlet 4 0% 20% 40% 60% Outlet 3 Dine In Orders 80% 100% 0% Revenue Per Square Foot By Outlet 20% 40% 60% 80% 100% Dine In Ticket Size by Outlet 3000 3000 Outlet 1 Outlet 2 2000 Outlet 3 1000 Outlet 4 Average Ticket (Rupees) 4000 Revenue Per Sq Ft (Ruppees) Outlet 4 Cumulative 0 Outlet Outlet 1 2000 Outlet 2 Outlet 3 1000 Outlet 4 Cumulative 0 Outlet A selection of results showing KPIs by outlet. Note that Outlet 1 leads in terms of revenue but when taking into account the square footage it is no longer the best performing. For the next step we narrowed our focus to the Dine-In channel 6 Selected Results –Part III Dine-In Channel Study The findings indicate that ‘Dine-In’ (DI) is a channel that could use some focus as it has the highest average ticket and could thus increase revenues. Across all outlets DI accounts for about 1/3rd of the revenue Order by Channel: Revenue & Orders Share Revenue Share Dine In Delivery Take Away Order Share 0% 20% 40% 60% 80% 100% Then, we focused on Outlet 1 and Outlet 2. The DI channel at Outlet 1 is fairly ‘healthy’, while at the latter the DI is lagging behind the other channels in terms of orders and revenue. Share of Orders by Channel: Outlets 1 & 2 Outlet 2 Share of Revenue by Channel: Outlets 1 & 2 Outlet 2 Dine In Dine In Delivery Delivery Take Away Outlet 1 0% 20% 40% 60% 80% 100% Take Away Outlet 1 0% 20% 40% 60% 80% 100% We look to understand and target customers better by looking at the DI purchase patterns using cluster analysis. 7 Cluster Analysis Background Data Selection and Methodology o Random selection of DI orders from Outlet 1 and Outlet 2. o There are 19 categories of items on offer. After the preliminary analysis we were able to reduce this number to 10. The categories (in alphabetical order) are 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. Beverage Chicken Combo Meals Dessert Noodles Prawns Rice Soup Squid Vegetables o Customers purchasing items from the same category have higher affinity to one another, similar nonpurchases do not carry as much ‘weight’. o Three different clustering techniques were attempted: spherical k-means r-Neighbourhood k Nearest Neighbour (kNN) o The best results were from the kNN-clustering method. The values chosen for k were k = 30 and k = 50. The k = 30 results are presented next 8 Selected Results – Part IV Cluster Analysis and Statistics The clusters are characterized by their purchases (listed in descending order). Interestingly the clustering was able to pick out the non-purchase categories as well. The clusters are: 1.Beverage/Squid/Soup/Rice/Chicken/Noodles 2.Combo 3.Rice/Beverage/Vegetable/Prawn/Chicken 4.Dessert/Rice/Beverage/Chicken/Squid 5.Squid/Rice/Beverage (No Dessert) 6.Rice/Beverage (No Soup/Vegetable/Squid) 7.Soup/Rice/Beverage (No Chicken) 8.Rice/Vegetables/Beverage (No Prawn/Squid) Cluster Percentage Contribution to Revenue Outlet 2 Outlet 1 Combined 0% 20% 1 40% 2 3 4 60% 5 6 7 80% 100% 8 Cluster Size by Percentage Outlet 2 Visiting Times of Top 4 Revenue Clusters Combined 0% 20% 1 40% 2 3 4 60% 5 6 7 80% 8 100% Cluster Number Outlet 1 6 4 3 1 0% 20% 40% Open - 3pm 3-6pm 60% 6-8pm 80% 100% 8pm - Close 9 Insights and Recommendations Based on the results of the Cluster Analysis • • • • Menu Optimization Set-Menu Value based options based on high frequency/revenue clusters (e.g. Cluster 3 Menu - Choose a Rice item, a Vegetable item, a Prawn item, a Chicken item and a beverage for Rupees XX) Time-based offers (e.g. Offer the ‘Cluster 3’ Set-Menu between 7 and 8 pm) Smart Website Recommendations When an item is selected, use the cluster analysis results to make recommendations – e.g. ‘Goes well with XXXX’ Loyalty Program (LP) Offers With a LP in place and customer purchase history, restaurant would be in a position to make individual/mass offers based on cluster 10 Argyle X Engagement Process Timeline for Project Manage & monitor Run Set up Week 1 Data Inventory + Q&A Analytics Solution for Business Problem Week 2 Week 3 Week 4 Week 5 Week 6 Data Selection Model Concept: KPI & Monitoring Metrics Data Integration + Data Validation Modelling and Statistical Analysis Model Validation Weekly deliverable review Issue capture, prioritization & resolution Customer Relationship Management Results & Findings Contact Us Head Office, Singapore: 133 Cecil Street #11-02, Keck Seng Tower 069535, Singapore Tel: +65 (82) 334424 Delivery Center, Sri Lanka: 37 Bagatalle Road, Colombo 3, Sri Lanka Tel: +94 (115) 666 111 Web www.argylex.com E-Mail [email protected] Linked In https://gr.linkedin.com/company/argylex Data driven insight. Insight driven outcomes. 12
© Copyright 2026 Paperzz