Advertising reports: denominators, zero conversions and small samples

THE RESULT

Check the input, then compare the complete deliverables

Explore the original materials, item-by-item outcomes and full report below. Attempt the input independently before using the reference to find omissions or disagreements.

Demonstration scope

Amounts are synthetic CNY values. No real Ads account or attribution report was read; no ad, bid or budget was changed.

Advertising reports practice

3 campaigns · 4 derived metrics

Original synthetic material. Try the input independently, then compare the reference results.

Original synthetic input

{
  "synthetic": true,
  "case": "google-ads-scoreboard",
  "data": {
    "currency": "CNY",
    "period": "2026-10-05/2026-10-12, end exclusive",
    "cvr_denominator": "clicks, by exercise convention",
    "conversions_threshold": 10,
    "campaigns": [
      {
        "id": "AD01",
        "spend": 240,
        "impressions": 6000,
        "clicks": 120,
        "conversions": 12,
        "conversion_value": 600,
        "previous_spend": 220,
        "previous_conversions": 11
      },
      {
        "id": "AD02",
        "spend": 80,
        "impressions": 2000,
        "clicks": 40,
        "conversions": 0,
        "conversion_value": 0
      },
      {
        "id": "AD03",
        "spend": 0,
        "impressions": 0,
        "clicks": 0,
        "conversions": 0,
        "conversion_value": 0
      }
    ]
  }
}

Processing rules

CTR=clicks/impressions; CVR=conversions/clicks; CPA=spend/conversions; ROAS=conversion_value/spend. A zero denominator means not calculated. Fewer than 10 conversions is a small sample, not a scale/stop decision. All amounts are CNY major units; verify actual API units and attribution separately.

Complete reference results under this tutorial’s rules; nothing was written to a third-party service.Scroll horizontally within the table if needed.

Campaign	CTR / CVR	CPA / ROAS	Interpretation
AD01	2% / 10%	CNY 20 / 2.5	12 conversions; prior CPA also 20; continue observation
AD02	2% / 0%	Not calculated / 0	0 conversions; small sample; investigate tracking/window
AD03	Not calculated / not calculated	Not calculated / not calculated	All denominators zero; no invented zero-cost performance

A comparable changeAD01

240/12=20 and 220/11=20. Increased total spend does not imply increased CPA.

The meaning of zeroCheck denominator

AD02 ROAS is 0/80=0, but CPA cannot divide by 0. AD03 has no calculable ratios.

Ads report — synthetic data
AD01: CTR 2%, CVR 10%, CPA CNY 20, ROAS2.5; prior CPA20, unchanged. No proof it can scale.
AD02: CTR2%, CVR0%, CPA not calculated, ROAS0; small sample; inspect tracking and attribution window.
AD03: no traffic/spend; all four ratios not calculated.
Missing: other baselines, real conversion actions and delayed attribution.
Suggestions are for human investigation. No campaign action.

This shows complete inputs and expected outputs. Switching tabs runs no task; verify real connections and writes separately in your environment.

BEFORE YOU START

Prepare the materials and check your tools

Start with the practice files

Download inputs and formulas.
Values are yuan, not cost_micros.
Only AD01 has prior spend/conversions; do not fill other baselines.

Before using real services

Verify customer account, currency, time zone, conversion action and attribution window.
Use one complete read-only period with fixed grouping.
Confirm the live CVR denominator instead of assuming the exercise definition.

Complete one file-based exercise first. For real services, verify the actual account, available tools and operation scope, then follow the acceptance check at each stage.

Download the original input and rules ↓

EXECUTION & TAKEOVER

Execution and takeover boundaries

The actions below require the agreed inputs, accounts and scope. Configuring a Routine does not expand that authorization.

Can run automatically

Read supplied ad reports; check periods, currencies and denominators, calculate supported metrics and flag small samples.

Requires confirmation

Confirm live account read and report delivery scope. Budget, bid, campaign pause and conversion-setting changes require separate approval.

Take over after a failure

Stop conclusions and account changes when units, attribution or data are unclear. Have a human compare the source report; keep zero-denominator metrics unavailable and recalculate.

STEP 01

Agree period, currency and conversion definitions

Similar metric names can use different events and denominators. Source units matter, especially currency micros versus major units.

Stage 01 · Agree period, currency and conversion definitions

Restate period, CNY units, CVR denominator and conversion threshold from source-data.json. List missing baseline/attribution information. Do not connect Ads or change budgets. Save metric-contract.md.

Expected output

A metric contract.

Acceptance check

Use yuan without dividing by a million; label the CVR convention.

STEP 02

Validate inputs and calculability

Check missing values and denominators before arithmetic. Zero conversions is not enough evidence to diagnose a tracking failure.

Stage 02 · Validate inputs and calculability

Check spend, impressions, clicks, conversions and value for each campaign. List numerator, denominator and calculation status for every metric. Do not fill missing values with zero or propose scale/stop actions.

Expected output

An input-quality table.

Acceptance check

Zero denominators in AD02/AD03 are explicit.

STEP 03

Calculate ratios with reasons for nulls

Not calculated is a legitimate result that prevents meaningless division. Keep formulas beside formatted display values.

Stage 03 · Calculate ratios with reasons for nulls

Compute all four ratios with formulas. Return null plus reason when the denominator is zero. AD02 ROAS is 0, CPA is null. Compare prior CPA for AD01 only; do not fill other baselines.

Expected output

Three reference metric rows.

Acceptance check

AD01: 2%, 10%, 20, 2.5; no Infinity, NaN or false zero CPA.

STEP 04

Write observations before investigation ideas

The threshold is an exercise convention, not proof of statistical significance. Small samples need investigation, not automatic budget changes.

Stage 04 · Write observations before investigation ideas

Write scope, metrics, comparable baselines, small-sample flags, missing information and human investigation suggestions. Give no scale/stop decision for AD02/03. State evidence and limits for every suggestion; do not assert causes.

Expected output

A report and investigation list.

Acceptance check

Spend changes are not directly called performance changes; sample flags remain visible.

STEP 05

Read a small live report and verify source fields

Delayed conversions can revise old totals. Link every summary to account, period, source field and observation time.

Stage 05 · Read a small live report and verify source fields

Read only <account/period/grouping>, at most 5 groups. Record fields/units, conversion action, time zone and attribution window. Recalculate one group from source values. Deliver files only; change no ads and send no Slack messages.

Expected output

A traceable live comparison or blocker.

Acceptance check

No bid/budget action; account, window and units can be verified.

STEP 06

Version backfills and definition changes

Late conversions require a dated revision. Different metric definitions should not silently replace historical numbers.

Stage 06 · Version backfills and definition changes

Draft an ads-scoreboard Skill with field/formula versions, period/attribution, zero denominators, small samples, backfill revisions and read-only scope. Propose a private-report Routine only. Review budgets and external delivery separately.

Expected output

A report and revision standard.

Acceptance check

Historical and revised definitions remain distinguishable.

TROUBLESHOOTING & ACCEPTANCE

Find the evidence before choosing the next action

CPA is infinite

Use not calculated for zero conversions and retain spend/reason.

Money is off by a million

Check the actual source unit rather than guessing from magnitude.

Reports disagree on conversions

Compare account, time zone, event, window and observation time.

Can small samples still support advice?

They can support investigation steps, not confident automatic budget actions.

Check every item before finishing

These page-local checkboxes reset on refresh. Checking them does not verify a real service connection.

Use yuan without dividing by a million; label the CVR convention.Zero denominators in AD02/AD03 are explicit.AD01: 2%, 10%, 20, 2.5; no Infinity, NaN or false zero CPA.Spend changes are not directly called performance changes; sample flags remain visible.No bid/budget action; account, window and units can be verified.Historical and revised definitions remain distinguishable.

Checked 0 / 6 items

PRACTICE MATERIALS

Download the inputs separately from the reference answers

Start with input and rules without showing the reference answers. After acceptance, save the report and complete prompt set.

Original practice inputInputs only; attempt the exercise before opening the reference answers.
source-data.jsonRules and scopeDefinitions, exceptions and delivery requirements.
workflow-rules.mdItem-by-item reference resultsCompare every input with its outcome.
expected-results.csvComplete reference reportResults, evidence and missing information.
reference-report.mdAll stage prompts6 instructions with expected outputs and acceptance checks.
practice-prompts.txt

SOURCES & VERIFICATION

Check documented capabilities and exercise results separately

Content updated: 2026-10-18 · Editing a page does not reverify its sources or live workflow.

Documentation review: Unverified · no review date recorded
No case-specific documentation review is recorded. References below are provided for review and do not establish verification.

Live workflow verification: Unverified
Live accounts, external writes, scheduled execution and long-term operation are unverified. Synthetic exercises and page checkboxes do not establish live verification.

Amounts are synthetic CNY values. No real Ads account or attribution report was read; no ad, bid or budget was changed.

Google Ads: Metrics ↗
Check source units and conversion definitions; formulas in this exercise use the stated metric contract.
Grok Bot: apps and the shared computer ↗
Check connected accounts, available tools, visibility and user takeover.
Grok Bot: Skills and Routines ↗
Validate a manual run before saving a method or scheduling it.

NEXT STEP

Try another small batch of your own material

Retain original input and operation records. Verify that each result can be traced before adding connections or recurring runs.

Return to Content & growth →

Learning path

Get started

Advanced

Recipes

Resources

Advertising reports: denominators, zero conversions and small samples

Check the input, then compare the complete deliverables

Prepare the materials and check your tools

Start with the practice files

Before using real services

Execution and takeover boundaries

Can run automatically

Requires confirmation

Take over after a failure

Agree period, currency and conversion definitions

Validate inputs and calculability

Calculate ratios with reasons for nulls

Write observations before investigation ideas

Read a small live report and verify source fields

Version backfills and definition changes

Find the evidence before choosing the next action

Check every item before finishing

Download the inputs separately from the reference answers

Check documented capabilities and exercise results separately

Try another small batch of your own material