sidebar.joinCommunitysidebar.communicateWithUsers
sidebar.joinAskTableCommunity
sidebar.wechat
sidebar.feishu
sidebar.chooseYourWayToJoin
sidebar.contactUssidebar.getProfessionalSupport
sidebar.contactUs
sidebar.scanToAddConsultant
sidebar.wechat
sidebar.feishu
sidebar.chooseYourWayToJoin
sidebar.scanToAddConsultant
"Sales dropped."
This sentence alone has no value. What's valuable is the chain of questions that follow:
The core difference between a senior analyst and a novice is whether they can systematically answer this chain of "which" questions.
AskTable's Drill-Down Metrics Skill does one thing: Transforms this "peeling the onion" analytical capability into an automated process—no need to try dimensions one by one, it automatically finds the clue with the largest contribution to difference, and follows it all the way to specific actionable problem points.
Drill-down analysis sounds simple—just break it down by different dimensions, right? But in practice, business staff often encounter three problems:
Problem 1: Too many dimensions, don't know which to look at
A sales dataset can be broken down by these dimensions:
- Time: day/week/month/quarter/year
- Region: province/city/district/store
- Category: major/middle/minor/SKU
- Channel: online/offline/direct/franchise
- Person: sales team/individual
- Customer: new/repeat/dormant
Combinations number in the hundreds/thousands. Where to start?
Problem 2: Looked at many dimensions, but can't find the key
You might discover "East China down 10%," but East China has 50 stores, 20 categories, 3 channels—which is dragging it down? Continue breaking down, find new leads. After several layers, information explodes, more you look, more confused you get.
Problem 3: Found the difference, but can't find action points
"East China Store A Category B Channel dropped"—so what? This conclusion can't directly translate to action. Good drill-down should ultimately point to specific problems that can be intervened.
AskTable's Drill-Down Metrics Skill follows three core principles:
Principle 1: Contribution priority
Not all dimensions are examined, but prioritize dimensions with largest contribution to difference.
(If East China contributed 80% of total decline, then focus on East China first)
Principle 2: Layer-by-layer progression
Each drill-down layer based on previous layer's findings, not parallel expansion.
(Found East China anomaly → drill down East China stores → found Store A anomaly → drill down Store A categories)
Principle 3: Until actionable
Drill-down doesn't stop at "found problem," but finds specific points where action can be taken.
(Not "East China dropped," but "East China Store A's X category stockout caused lost sales")
When you ask "Sales dropped, which area is dragging it down?", AskTable:
Step 1: Calculate each dimension's contribution
Total sales decline: -1 million
By region contribution ranking:
1. East China: -650,000 (contributed 65%) ← Core driver
2. South China: -200,000 (contributed 20%)
3. North China: -100,000 (contributed 10%)
4. West China: -50,000 (contributed 5%)
By category contribution ranking:
1. 3C Digital: -550,000 (contributed 55%) ← Core driver
2. Home Living: -250,000 (contributed 25%)
3. Food Beverage: -120,000 (contributed 12%)
4. Other: -80,000 (contributed 8%)
Step 2: Cross-locate core problem
Region × Category cross analysis:
- East China × 3C Digital: -420,000 (42% of total decline) ← Core problem
- East China × Home Living: -150,000
- South China × 3C Digital: -100,000
AskTable automatically finds "largest contribution intersection point," rather than making you manually try dimensions one by one.
AskTable doesn't just look at dimensions in parallel—it constructs a drill-down path:
This path's value: Each step tells you why you're looking at this dimension, and what this dimension revealed.
Drill-down doesn't proceed infinitely. AskTable auto-stops in these situations:
| Stop Condition | Description |
|---|---|
| Found actionable problem | e.g., "SKU stockout," "store system failure" |
| Difference no longer significant | After continued drill-down, all sub-items have small differences—sufficient granularity reached |
| Data granularity at limit | Already at finest granularity (single SKU, single order) |
| Contribution too low | Continued drill-down dimension contribution < 5%, not worth continued investment |
User asks: "Sales this month 1 million less than last month, help me see where the problem is."
📊 Drill-Down Analysis Report
Level 1 breakdown (by region):
- East China: -650,000 (65%) ← Core
- South China: -200,000 (20%)
- Other regions: -150,000 (15%)
Level 2 breakdown (East China by store):
- Store A: -390,000 (60% of East China) ← Core
- Store B: -150,000
- Other stores: -110,000
Level 3 breakdown (Store A by category):
- 3C Digital: -270,000 (69% of Store A) ← Core
- Home Living: -80,000
- Other categories: -40,000
Level 4 breakdown (3C Digital by SKU):
- Model X phone: -180,000 ← Root cause
- Model Y tablet: -60,000
- Other SKUs: -30,000
🔍 Root Cause Location:
Model X phone stockout from 15th of month, continuous 15 days out of stock.
This SKU normally sells 12,000/day, 15-day impact ~180,000.
✅ Action Suggestions:
1. Emergency replenish Model X phone
2. Check if Store A's auto-replenishment rules failed
3. Investigate if other stores have same potential issue
User asks: "New users this month decreased, where's the problem?"
Level 1 breakdown (by channel):
- Organic search: -30% (contributed 45% of total reduction) ← Core
- Paid ads: -10% (contributed 30%)
- Social media: -5% (contributed 25%)
Level 2 breakdown (organic search by keyword):
- Core brand keywords: Traffic normal
- Industry generic keywords: Traffic down 50% ← Core
- Long-tail keywords: Traffic down 20%
🔍 Root Cause Location:
Industry generic keyword ranking dropped from #2 to #8,
mainly affected by competitors increasing SEO investment.
User asks: "This month's costs exceeded budget by 15%, help me break down reasons."
Level 1 breakdown (by cost type):
- Raw material cost: Over budget 1.2M (contributed 60%) ← Core
- Labor cost: Over budget 500,000 (contributed 25%)
- Logistics cost: Over budget 300,000 (contributed 15%)
Level 2 breakdown (raw materials by category):
- Chips: Over budget 700,000 (price increase 18%) ← Core
- Displays: Over budget 300,000 (price increase 8%)
- Other materials: Over budget 200,000
🔍 Root Cause Location:
Chip prices increased 18%, mainly due to global supply chain tightness.
Current inventory can sustain 2 weeks, recommend locking prices ASAP.
"Sales dropped, which area is dragging it down?"
"Which region/category/time period has the problem?"
"Help me break down this month's cost overrun reasons"
"Where did the new user decrease mainly come from?"
If you have special interest in a particular dimension:
"Break down East China sales by store dimension"
"Show me this month's sales trend by day"
You can also let AskTable complete multi-layer drill-down automatically:
"Help me go through it from start to finish, find the core problem"
"Do a comprehensive analysis, see which areas have problems"
AskTable automatically constructs drill-down path, all the way to specific actionable problem points.
Drill-down is the core skill for diagnosing problems, typically linking with:
Anomaly Detection (discovered problem: metric anomalous)
↓
Drill-Down Metrics (locate scope: where the problem is)
↓
Attribution Analysis (quantify cause: each factor's contribution)
↓
Metric Interpretation (business translation: what it means)
↓
Report Orchestration (output results: complete diagnostic report)
Drill-down's value: Transforms the vague question "where did it go wrong" into a clear diagnostic path, ultimately pointing to actionable specific problem points.
Pain point: Each sales anomaly, operations team spent 3+ hours opening various reports, checking region, category, channel one by one to find root cause. Easily missed key dimensions during investigation.
Solution: Enable Drill-Down Metrics Skill, link with sales data dashboards.
Effects:
"Before when problems happened, whole team went to check data, took most of the day to find causes. Now AskTable automatically breaks it down for me, directly tells me where the problem is and what to do. Time saved, we can focus on strategy optimization." —— Operations Manager, a certain e-commerce company
Drill-Down Metrics Skill's core value isn't in "how many layers you can break down," but in:
The ultimate state of drill-down analysis isn't breaking open all data to look through, but using the fewest breakdown steps to find the most worthwhile problems to solve.
sidebar.noProgrammingNeeded
sidebar.startFreeTrial