VI
Diagnostic analysis · Diagnostic Analytics

From the number,
find the problem.

The figures tell you what is happening — but they rarely tell you why on their own. Each framework below is a way to pull things apart: it feeds on one specific kind of number, and surfaces one specific kind of problem. Grouped by the source of the input data.

Revenue · 12 periods▼ red flag
The top-line number climbs, then breaks. The framework tells you where it broke, because of whom, and where you're still bleeding.
01 · Internal
Numbers generated by the business itself: financial books, sales, customers, operations, inventory.
02 · External
Numbers from the market: industry benchmarks, competitors, market share, market size, macro variables.
03 · Synthesis
Combining multiple sources and bridging from symptom to cause — the reasoning part, not something a machine does on its own.
GROUP 01

Internal data

Input · the business's own books Most of the work of "reading a number into a problem" lives here. These frameworks decompose the top-line metric, examine it over time and break it down by segment — so the problem isn't hidden by being "averaged away".

01.1

DuPont Analysis

ROE decomposition
InputFinancial statements: profit, revenue, assets, equity.
MethodSplit ROE = profit margin × asset turnover × leverage.
SurfacesWhether weak return on equity comes from thin margins, sluggish assets, or too much borrowing.
01.2

Variance Analysis

Deviation analysis
InputActuals compared against plan / budget / prior period.
MethodSplit the variance into a price component and a volume component.
SurfacesWhether missing target is because you sold less, or because you sold cheaper.
01.3

Revenue Bridge

Bridge / waterfall chart
InputRevenue across two periods, split by source of increase/decrease.
MethodBuild the steps: new customers, lost customers, price increases, order expansion.
SurfacesWhat's really driving revenue, and where the leaks are.
01.4

Driver Tree

Factor tree
InputOne KPI and the sub-metrics that make it up by formula.
MethodDecompose by branch: traffic × conversion rate × order value.
SurfacesWhich branch of the tree is off and dragging the whole KPI down.
01.5

Cohort Analysis

Grouping by join date
InputCustomers tagged by the date they joined.
MethodTrack each cohort's behavior across its lifetime.
SurfacesWhether newer customers retain worse than older ones — quality is slipping.
01.6

Segmentation

Breaking into segments
InputNumbers split by channel, product, region, customer group.
MethodSlice the top-line number to expose the anomalous segment.
SurfacesA problem hidden in one specific slice that the total flattens out.
01.7

RFM Analysis

Recency · Frequency · Monetary
InputTransaction history: most recent purchase, frequency, value.
MethodScore and classify customers along three axes.
SurfacesHigh-value customers going cold, and which groups are worth keeping.
01.8

Funnel Analysis

Conversion funnel
InputCounts through each step: visit → sign-up → purchase → return.
MethodMeasure the drop-off rate between consecutive steps.
SurfacesWhich step the most customers abandon at.
01.9

Pareto Analysis

The 80/20 principle
InputValue distribution by product / customer / failure cause.
MethodRank and find the minority that produces most of the impact.
SurfacesWhich few causes account for most of the problem.
01.10

Unit Economics

Economics per unit
InputCustomer acquisition cost (CAC), lifetime value (LTV), contribution margin.
MethodCompute profit/loss on a single customer or a single order.
SurfacesLosing more the more you sell — growth is burning cash.
01.11

Ratio Analysis

Financial ratios
InputLine items from the financial statements.
MethodA set of liquidity, leverage, profitability and efficiency ratios.
SurfacesA ratio off the benchmark / off its history is a red flag to investigate.
01.12

Break-even Analysis

Break-even point
InputFixed costs, variable costs, selling price.
MethodFind the volume/revenue to break even, and test the sensitivity.
SurfacesHow much you need to sell to turn a profit, and how far you are from that line.
01.13

Cash Conversion Cycle

Cash cycle
InputInventory, accounts receivable, accounts payable.
MethodMeasure the number of days cash is stuck inside operations.
SurfacesProfitable on paper, but cash is jammed in inventory and receivables.
01.14

Trend & YoY / MoM

Internal time series
InputThe same metric measured repeatedly across many periods.
MethodCompare against the same period last year / the adjacent period to strip out seasonality.
SurfacesTrend break points and the real momentum once seasonality is removed.
01.15

Control Chart

Control chart · SPC
InputOperational / quality metrics measured continuously.
MethodDraw the bounds of "normal" variation around the mean.
SurfacesWhich swings are noise and which are genuinely abnormal.
GROUP 02

External data

Input · market, competitors, macro Internal numbers alone can't tell you "is bad actually bad, or just the nature of the industry". These frameworks place the business next to its external context, to tell whether the problem is yours alone or the whole market's.

02.1

Benchmarking

Against the industry standard
InputYour metrics placed next to the industry median / benchmark.
MethodCompare relative to the general baseline.
SurfacesWhere you're below par — margin, cost, or productivity.
02.2

Market Share Analysis

Market share analysis
InputYour revenue and the size of the market / competitors.
MethodCompute market share and its relative momentum.
SurfacesYou're growing slower than the market — losing your foothold.
02.3

Market Sizing

TAM · SAM · SOM
InputTotal market size, the serviceable segment, the obtainable share.
MethodEstimate the market ceiling in layers.
SurfacesWhether growth stalled because you hit the ceiling or because you're under-penetrating.
02.4

Macro Correlation

Correlation with macro variables
InputInternal results and macro variables (interest rates, input prices, season).
MethodMeasure correlation with exogenous factors beyond your control.
SurfacesWhether the decline is the broad market or something internal to the business.
02.5

Competitive Trend

Industry & competitor trends
InputPrice, product and growth figures for competitors and the industry.
MethodTrack relative movement over time.
SurfacesWhether your problem is unique to you or an industry-wide headwind.
02.6

Price Benchmarking

Price comparison & elasticity
InputYour price placed next to market prices and the demand response.
MethodCompare the price level, estimate the price elasticity of demand.
SurfacesLosing customers because your pricing is off the market — too high or too low.
GROUP 03

Synthesis analysis

Input · multiple sources · symptom → cause This is the bridging part. These frameworks combine internal signals with external ones, quantify the relationships, and trace from symptom down to root cause. Mostly the analyst's reasoning — the machine only assists.

03.1

Correlation / Regression

Correlation · regression
InputAn outcome variable and many driver variables, internal and external.
MethodQuantify the size and direction of each factor's effect.
SurfacesWhich factors truly drive the outcome, and which are just coincidence.
03.2

Sensitivity Analysis

Sensitivity analysis
InputA model with several input assumptions.
MethodChange one assumption at a time and measure the output's response.
SurfacesWhich variable the result is most fragile to.
03.3

Attribution Analysis

Contribution attribution
InputThe total result and the channels / factors that jointly drive it.
MethodAssign each factor's share of the overall result.
SurfacesWhich channel truly creates value, and which just rides along.
03.4

Anomaly Detection

Anomaly detection
InputMeasured metrics against an expected baseline (internal + market).
MethodCatch points outside the bounds using statistics / algorithms.
SurfacesSudden break points worth a deep look, surfaced automatically.
03.5

5 Whys

Asking "why" repeatedly
InputA symptom already isolated from the numbers.
MethodAsk "why" several times to trace down to the root.
SurfacesThe root cause sitting beneath the surface symptom.
03.6

Fishbone · Ishikawa

Fishbone diagram
InputThe problem and its plausible cause categories.
MethodGroup causes by category: people, process, tools, materials.
SurfacesA full map of cause directions so none get missed.
03.7

Hypothesis-driven

MECE · consulting style
InputHypotheses about the cause + data to test them.
MethodFrame hypotheses that don't overlap and don't leave gaps; use data to confirm/reject.
SurfacesRules out wrong hypotheses fast and locks onto the cause that holds up.
03.8

Scenario Analysis

Scenario analysis
InputA model + combinations of best / base / worst assumptions.
MethodRun multiple scenarios and compare the outputs.
SurfacesLatent problems that emerge under adverse conditions.

You rarely use a single framework alone

In practice they get chained together. Start by isolating which number is broken, find the segment causing it, and only then lock in the cause.

STEP 01

Decompose

Group 01 — break down the top-line metric to isolate which number is broken.

STEP 02

Set the context

Group 02 — compare against the industry and competitors to tell whether the problem is yours or the market's.

STEP 03

Break down the segment

Cohort & segmentation — pinpoint exactly which segment or step is creating the problem.

STEP 04

Lock in the cause

Group 03 — 5 Whys / hypothesis to confirm the root cause and rule out the wrong ones.