Português
EnglishQwenWork Practical Guide for Restaurant Owners
QwenWork Operating Guide
Running a restaurant involves more than just preparing dishes. Before service hours, you need to check if all outlets are operating normally; during downtime, you must respond to negative reviews and calculate profits. When planning new locations, you have to scout sites and oversee renovations. Information is scattered across various systems, and even after spending time reviewing everything, it can be hard to identify the most urgent tasks.
This article summarizes three QwenWork applications for the food industry: the Takeout Management Assistant for takeout operations, the Site Selection Advisor using Amap Cloud Map to evaluate locations, and the KeRuyun Restaurant Operations Analysis Assistant for monitoring sales and business conditions. We recommend selecting the specific problems you face and addressing them one by one—there’s no need to install all features at once.
QwenWork serves as the entry point, while expert suites handle specific tasks. To access your authorized data, you must connect your store accounts.
Step 1: Activate Tools and Connect Your Restaurants
Open qwenwork.cn/app on your computer and log in with your phone number or by scanning the DingTalk QR code. The web version doesn’t require installation; for the desktop app, download it from the “Download Desktop Version” link on the QwenWork website. DingTalk users can also update to the latest version and access QwenWork from the left sidebar.
The expert suite provides the tools, while connectors access other systems within your authorization scope. After installing the suite, complete the connection and authorization steps. For your first use, select a familiar store and review a single day’s data to confirm that store names and numbers are correct before expanding.
For takeout: Connect “Taobao Flash Food Merchant Edition” and “Browser Automation,” then install and enable the “Takeout Management Assistant.”
For Amap: Connect “Amap Cloud Map” and “Browser Automation,” then install and enable the “Site Selection Advisor.”
For KeRuyun: Enable “Browser Automation,” install the “KeRuyun Restaurant Operations Analysis Assistant,” and log in with an account that has the necessary permissions. When combining Taobao data, always use it alongside the Takeout Management Assistant.
Managing Takeout: 14 Scenarios to Choose From
All scenarios below use the “Takeout Management Assistant.” Replace the bracketed terms with your store name, date, product, or group name.
01 Business Hours Have Arrived—Which Stores Aren’t Active Yet?
The kitchen is ready, staff are on duty, but the takeout platform hasn’t signaled that the store is open.
Prompt: Check the operating status of [store range] for today. Based on each store’s schedule, identify those whose business hours have started but remain closed, excluding normal breaks. List store names, cities, current status, and duration of the issue. If the start time of the closure is unknown, note this explicitly—do not estimate. Just list the stores for now.
Your priority is ensuring customers can place orders; boosting sales comes second.
02 Why Do Different Stores Under the Same Brand Have Varying Minimum Orders and Delivery Fees?
A customer may abandon their order if they’re still far from meeting the minimum purchase requirement when checking out.
Prompt: Compare the operating hours, minimum order amounts, delivery fees, and free shipping thresholds of [store range], using [headquarters standards] as the benchmark. If headquarters standards aren’t available, reference the settings of most stores and note this. Output stores with discrepancies, along with their current values, comparison values, and items that differ. Just check for issues—don’t make changes yet.
Clarify differences before deciding which settings need standardization.
03 There Are Many Negative Reviews—Which Should You Address First Today?
Owners worry about two things: serious complaints hidden among many reviews, and ordinary negative feedback left unanswered for days.
Prompt: Retrieve the last 30 days’ new negative reviews for [store range], including those that remain unresponded to. Categorize by store, rating, and issue type, listing the review content, time, and response status. Provide a prioritized list of reviews to address. Place food safety concerns at the top and mark them as pending verification. Just review—do not reply automatically.
The goal is to identify issues that need immediate attention.
04 Too Many Unanswered Reviews—How Can You Handle Them One by One?
The lunch rush has just ended, and a new batch of reviews has piled up.
Prompt: Filter all unresponded negative reviews for [store range], categorizing them by food quality, missing items, wrong deliveries, delivery times, packaging, and promotions. Write individual responses based on the original text. Separate food safety feedback and mark it as pending offline investigation. Do not make false compensation promises. Show me a preview first; only proceed with bulk replies after my approval, and output the results.
You can handle multiple reviews at once, but don’t treat every customer’s problem the same way.
05 Sales Fell This Week—Which Metrics Should You Focus On First?
Daily sales figures only tell you whether business is good or bad.
Prompt: Generate a weekly operations report for [brand or store], specifying the reporting period (up to yesterday, excluding today’s incomplete data). Summarize sales, revenue, order volume, impressions, in-store conversion rate, online-to-in-store conversion rate, and average spend. Compare these metrics to the previous week, yesterday, and industry averages. Explain major changes, key areas to investigate, and any missing data in plain language.
The value of a weekly report is helping you choose one priority task for next week.
06 Feeling That Business Is Slowing Down—Are Customers Simply Not Buying?
Conversion rates sound complicated, but they simply measure what percentage of visitors take the next step.
Prompt: Analyze business trends for [store range] over [time period]. First, confirm whether the online-to-in-store conversion rate has indeed dropped. Then examine: Which cities, districts, and stores show the biggest declines? Which products are selling less? Have prices or promotions changed? Are there stockouts or slower deliveries? What are customers complaining about in reviews? Identify the main issues, relevant data, and recommended actions. Note any missing data; just analyze—don’t adjust operations yet.
Identify the root cause before deciding where to allocate resources.
07 Out-of-Stock Items Still Listed Online—What Should You Do?
If a dish is unavailable but still listed, you risk receiving orders you can’t fulfill. Conversely, if an item is back in stock but still marked as sold out, you lose potential sales.
Prompt: In [specific store], remove [items to be discontinued] from the menu and restore [restocked items] to appropriate inventory levels. First, list the store, item, variant, current status, target level, and any constraints. Show me the proposed changes. Only execute after my confirmation, and return details on success, failure, and unexecuted items.
The menu should reflect what the kitchen can actually prepare.
08 Need to Raise Prices Across Multiple Stores—How to Avoid Missing Any or Changing the Wrong Item?
If the same signature dish’s price is adjusted separately at different locations, some stores might miss the change, or others might alter the wrong variant.
Prompt: Uniformly increase the price of [specific item or combo] by [amount] yuan across [city and store range]. First, list the store, item, current price, and adjusted price. Also check for promotional conflicts and note any items that cannot be adjusted or are unclear. Proceed with bulk execution after my confirmation, outputting details on successful, failed, and unexecuted changes, and retaining backups of original prices.
Confirm the target items before adjusting prices, and verify the results afterward.
09 How to Quickly Understand Yesterday’s Business Before Opening?
The challenge: Owners check multiple dashboards daily but still struggle to explain what went well and what didn’t yesterday.
Prompt: At [time] every morning, summarize the previous day’s performance across authorized stores. Include sales, valid orders, average spend, and online-to-in-store conversion rates compared to the previous day and the same day last week. List impressions, in-store visits, online orders, and both conversion rates, highlighting weak links. Name the top-performing stores, the 1–2 worst performers, and note any significant drops in ratings, unresolved reviews, or star changes. Use simple, store-manager-friendly language. Generate a draft first; after DingTalk connector authorization and confirmation of the [receiving group name], send it as scheduled.
The purpose of the morning report is to decide which store to focus on and what action to take today.
10 How to Detect Order Backlogs, Refunds, and Delayed Deliveries Early?
When orders pile up, meals aren’t ready, and couriers crowd the store waiting to pick them up.
Prompt: At [check time] every day, review yesterday’s data across authorized stores and trigger alerts under any of these conditions: 1. Sales or order volume drop by more than 20% compared to the previous day or the same day last week. 2. Online-to-in-store conversion rate falls 3 percentage points or more below the 7-day average. 3. Refund rate reaches 10% or higher—stores with fewer than 20 orders are exempt. 4. Store rating falls below 4.5, or at least 3 new negative reviews appear. 5. Actual receipts fall below 70% of total payments. During operating hours, check for stores that should be open but aren’t, excluding holidays and planned closures; also flag late deliveries or orders expected to be late within 5 minutes. List actual numbers, triggering rules, and initial investigation directions by store. Prioritize food safety reports, widespread closures, and order backlogs; otherwise, report normal conditions. Provide a draft first; after DingTalk authorization and confirmation of the [receiving group name], send it regularly.
Alerts must include specific numbers, store names, and next steps to avoid becoming noise in the group.
11 Who’s Driving Weekly Sales Fluctuations?
The problem: Prices are cut when sales dip and promotions added when they rise, without understanding the real causes.
Prompt: Every Saturday, review the past two weeks. Default to analyzing online-to-in-store conversion rates, starting with in-store and online order volumes. If revenue fluctuates by more than 15%, focus on revenue instead. Break down contributions by impressions, in-store conversion, online conversion, and average spend, keeping definitions consistent. Trace the stores and factors with the greatest impact. Compile all negative reviews, separating merchant-side and delivery-side issues with typical examples. Examine subsidy effects and abnormal order/delivery rates on peak days. Suggest three priority actions for the following week. Provide a draft first; after DingTalk authorization and confirmation of the [receiving group name], send it regularly.
First find the biggest drivers of change, then plan next week’s moves—don’t cut prices indiscriminately whenever sales slip.
12 Customers Pay a Lot—Why Does the Store Receive Less?
The problem: Orders look strong, but the final payout is disappointing.
Prompt: At [time] every morning, reconcile total customer payments with actual store receipts across all authorized stores. Only count officially recorded “store receipts”; do not substitute other ambiguous amounts. Calculate the receipt-to-payment ratio for each store and the overall brand, comparing it to the previous day. Highlight stores with ratios below 70%; exclude those with fewer than 20 orders. List the top five orders with high payment amounts but low receipt ratios, showing how much was paid, how much received, and what was purchased. Every Wednesday, add checks on ongoing promotions, reduced shipping, targeted coupons, and platform invitation offers. Sample several large, low-ratio orders, detailing how much discount is covered by the platform, merchant, and coupons, and identifying stores bearing unusual burdens. Assess whether subsidies generated new orders, noting any lack of evidence. Just observe—do not modify promotions or prices, and do not disclose customer information. Retry once if data retrieval fails; otherwise, log it as uncovered and proceed with unaffected stores. Provide a draft first; after DingTalk authorization and confirmation of the [receiving group name], send it regularly.
First determine how much the store actually receives, then investigate who bears the discounts, and finally decide how to adjust promotions.
13 Which Menu Items Should Be Promoted More, and Which Need Adjustments?
The problem: The menu is long, but managers rely on intuition to choose which items to discount.
Prompt: Every morning, analyze impressions, in-store visits, online orders, and both conversion rates for authorized stores, comparing them to the 7-day average to identify weak links and lagging stores. Every Friday, examine menu items by impressions, clicks, click-through rates, sales volume, and sales revenue, categorizing them into four groups: Keep high-impression, high-conversion items; investigate high-impression, low-conversion items regarding images, pricing, and reviews; list high-conversion, low-impression items as potential stars and suggest increasing exposure; and flag low-impression, low-sales items as candidates for removal. Also list high-impression, low-click items and prioritize image optimization; for high-click, low-order items, focus on pricing and reviews. Just provide analysis and recommendations; generate a draft first, and send it after DingTalk authorization and confirmation of the [receiving group name].
First see where customers are getting stuck, then decide whether to promote, adjust, or monitor further.
14 Data Review Complete—What Should Managers Do Today?
The problem: Reports and alerts fill the group chat, but managers don’t know where to start.
Prompt: At [time] every morning, compile yesterday’s data into a “Manager’s Daily Action Plan” for authorized stores. Start with a summary of each store’s status, then assign tasks: 1. Priorities for improving platform health, ranked by platform weight, with official benchmarks. Mark unverified standards as pending. 2. Number of negative reviews and main complaints. Place food safety concerns at the top and handle them offline—no template replies. Mark delivery-related issues as carrier responsibility. 3. Factors dragging down ratings below 4.2; do not draw conclusions if fewer than five reviews exist, but still address serious feedback. 4. Cancelled, refunded, and delayed orders; zero sales, out-of-stock, or removed signature dishes; weaker links between in-store and online channels. Prioritize must-do tasks for today, followed by issues affecting this week’s conversion, and finish with minor chores. Attach data sources, specific actions, and completion criteria to each item. Provide a draft first; after DingTalk authorization and confirmation of the [receiving group name], send it regularly.
A good briefing lets managers act immediately and verify completion at closing.
Amap Site Selection: Compare Two Locations, Then Verify On-Site
Suppose you’re opening a noodle shop—one site near office buildings, the other near residential areas. Prepare the addresses, sizes, monthly rents, cuisine type, target customers, average spend, and budget beforehand. Simply asking “Help me find a good location” won’t yield useful comparisons.
The Site Selection Advisor analyzes five aspects: nearby foot traffic, whether it matches your target audience, competition from similar businesses, transportation convenience, and surrounding commercial amenities. Look beyond the score to understand the basis of each judgment.
Prompt: Use the “Site Selection Advisor” to compare [Address A] and [Address B]. I’m opening a [cuisine type] targeting [customer group], with an average spend of [amount]. Using Amap Cloud Map data, compare demographics, competition, transportation, and commercial facilities, then weigh pros and cons against rent. Note sources and dates; mark missing items as “to be supplemented” and list what needs on-site verification.
After receiving the report, visit both locations during lunch and dinner: observe where people come from, how easy the entrance is to find, and whether customers can enter comfortably. Add your observations to refine the analysis. Traffic estimates, conversion rates, and revenue projections based on assumptions should not be treated as guaranteed income.
Once you’ve chosen a location, you can use the Multi-Person Workbench to track renovations, permits, and expenses. For small shops, four categories suffice: progress, bottlenecks, permits, and spending. Clearly assign responsibilities, set deadlines, and record current statuses so owners can see who’s handling what and where delays occur.
Prompt: Use the “Site Selection Advisor” and the Multi-Person Workbench to create a pre-opening workbench for [new store]. Collect responsible parties, opening date, budget, and permit status, then organize progress, bottlenecks, permits, and spending. Remind me if information is missing—do not assume anything is complete.
The delay and cost overrun thresholds in the example can be adjusted to fit your situation. They are management rules, not industry standards. Progress updates in the workbench must be provided by designated individuals—don’t assume the table itself ensures accountability.
KeRuyun Daily Report: Combine Dine-In and Takeout, Align Numbers First
Focusing solely on takeout orders risks overlooking the other half of your business. The KeRuyun Restaurant Operations Analysis Assistant helps consolidate POS data; paired with the Takeout Management Assistant, it supplements Taobao traffic, conversion, and promotion analysis.
Start with a familiar store and a single day’s data. Have it list the store, reporting period, data sources, and missing items, and cross-check with your backend before scaling up. If a store is missing, check permissions first—don’t claim coverage for “all stores” prematurely.
Prompt: Use the “KeRuyun Restaurant Operations Analysis Assistant” and “Takeout Management Assistant” to analyze [store range] on [date]. First verify data sources and reporting scope to avoid double-counting takeout revenue. Summarize total operations, channel mix, store performance, promotion effectiveness, anomalies, item margins, and payment reconciliation, then suggest three priorities for today. Note any missing data; do not adjust prices, promotions, or financial records yet.
The report focuses on four questions: How much was sold yesterday? Which stores showed the biggest changes? Which items need attention? What should be done today? Small shops can use a simplified version; for larger chains or complex issues, add detailed takeout metrics, conversion rates, and repeat purchase analysis.
When integrating the two systems, avoid double-counting. KeRuyun may already include takeout revenue, so don’t simply add Taobao figures again. Clarify the meanings of sales, gross revenue, and store receipts; list uncertain amounts separately rather than forcing a total.
You also need reliable cost data for item margins. Don’t calculate profit by subtracting commissions from list prices. If a combo sells well but has low margins, first check ingredient costs and who covers the discounts before deciding whether to adjust. The 30% margin threshold in the example isn’t suitable for all restaurants.
If takeout orders decline but dine-in remains steady, investigate takeout channels first rather than applying blanket promotions. Clearly distinguish between supported findings and unverified speculations in the report’s conclusions.
Last Step: Get It Right Once, Then Automate Daily
Manually generate a report, verifying stores, timing, and data sources. Once satisfied, set it to run daily.
Prompt: Schedule the task to run every day at 8 AM Beijing time, covering the same stores as the previous day, saving results in a chat first.
To send it to managers, configure the DingTalk connector and specify the group name.
Run it once immediately to ensure the report generates and messages are delivered. For tasks involving local browser automation, keep your computer awake and connected. Treat missing data as absent, not zero, and speculative causes as unconfirmed.
You don’t need to run all functions every day. Tomorrow, before opening, pick the most pressing issue, copy the corresponding prompt into QwenWork, and test it once.
Let the system handle the backend; leave the customers to you.
DomTech is DingTalk’s official service provider in Macau, dedicated to providing DingTalk services to a wide range of customers. If you’d like to learn more about DingTalk platform applications, contact our online support directly, or call +852 95970612 or email [email protected]. With a skilled development and operations team and extensive market experience, we can offer professional DingTalk solutions and services!