Reader resources · worldwide · free · no signup

Companion resources for The Agency Reporting Playbook

The working references from the book, in one place: the 30-day checklist, the maturity model, a KPI dictionary starter, the architecture, the weekly checklist, and the build-vs-buy worksheet. Everything on this reader page is available worldwide and requires no email.

New here? About the book →

0130-Day Reporting System checklist

The install month, as a checklist.

The chapter 13 sequence - decide, plumb, model, present - broken into checkable steps. Print it, or copy it into your project tool.

Week 1

Decide

Write the KPI dictionary: one line per metric - definition, formula, source
Agree a single UTM naming convention and write it down
Choose the stack (connectors, warehouse, dashboard tool) and record why
Pick the two clients you will install first
Name one owner for the install month

Week 2

Plumb

Connect each source so raw data lands on a schedule - no hand exports
Verify data arrives for 3 consecutive days before building anything on it
Record source quirks you hit (timezones, attribution windows, delays)
Leave raw data raw - transformations belong to next week

Week 3

Model

Encode each KPI dictionary line as a modeled definition over the raw data
Reconcile modeled numbers against the platforms and document known gaps
Add a basic data-quality check on the numbers clients notice most
Have a second person verify one metric end to end

Week 4

Present

Build one templated dashboard reading only the modeled layer
Write the one-page report: headline, three numbers, commentary, next step
Onboard the remaining pilot client from the template - time it
Write the weekly reporting SOP while the steps are fresh

02Maturity model reference

The four stages, with their tell-tale signals.

Chapter 2 in reference form. Read the signals honestly - the stage you are at is the one whose signals make you wince.

Stage 01

Manual

→Reports take 3-6 hours to produce each week
→Every client deck is formatted differently
→Data lives in ad hoc spreadsheets with no version control
→One team member is the single point of failure

Stage 02

Fragile

→Templates exist but require manual cleanup each week
→UTM naming is inconsistent across team members
→GA4 and ad platform numbers frequently disagree
→New clients take 2-3 days to onboard into reporting

Stage 03

Scalable

→Dashboard templates are reused across all client accounts
→KPI definitions are documented and team-wide
→A new account is live in reporting within 2-4 hours
→Reporting SOP exists and is followed consistently

Stage 04

AI-ready

→Data flows automatically: GA4 → BigQuery → dashboard
→AI prompts generate accurate insight commentary on clean data
→Weekly report generation takes under 15 minutes
→Data quality is monitored and alerts fire before clients notice

Want a scored diagnosis instead of self-placement? Take the free 3-minute assessment →

03KPI dictionary starter

Eight metrics, defined once.

The starting rows for your own KPI dictionary (chapter 6). Copy them, then adapt the definitions to how YOUR agency counts - the value is that they are written down.

Metric	Definition	Common pitfall
Spend	Total advertising cost for the period, per platform, in account currency.	Currency mixing and partial-day data on the current day.
Impressions	Times an ad was served, as reported by each platform.	Not comparable across platforms - do not sum into one number without a note.
Clicks	Platform-reported ad clicks (not sessions).	Clicks and sessions will never match; decide which one the report leads with.
Sessions	Analytics-reported visits (GA4 definition).	GA4 sessions differ from old UA sessions and from platform clicks.
Conversions	The ONE action you count, named explicitly (e.g. purchase, qualified lead).	The most common source of client mistrust: each platform counts differently. Define it once, in writing.
Revenue	Transaction value attributed to the period, from the source of truth (store/CRM, not the ad platform).	Platform-reported revenue includes modeled attribution; store revenue is what the client reconciles against.
CPA	Spend ÷ conversions, using the definitions above.	Only meaningful once "conversions" is pinned down.
ROAS	Revenue ÷ spend, stating which revenue source is used.	Platform ROAS and blended ROAS are different metrics - label them.

Part V of the Kindle book includes the four worked dictionaries, a modeling project, data-quality tests, dashboard specifications, client-facing documents, and operating checklists. The companion material here is a quick-use reference, not a paywall around what the book already includes.

04Reporting architecture reference

The warehouse-first shape, at a glance.

Sources

Ingestion

Warehouse

Modeled data

Dashboards

AI insight

One rule holds it together: dashboards never read raw platform data directly. They read the modeled layer, where each KPI is defined exactly once - so two dashboards cannot disagree, and an AI tool reads the same clean truth your clients see.

05Weekly reporting checklist

The Friday routine, in six lines.

01Data check first: did every source load since the last report? If not, fix before writing.
02Scan the numbers clients notice most (spend, conversions, revenue) for anomalies.
03Write the 90-second version first: headline, three numbers, one recommended action.
04Commentary answers "so what" - what changed, why, what we are doing about it.
05Send on the agreed day, every week, even when the news is bad. Consistency is the trust.
06Log anything you fixed by hand - two hand-fixes of the same thing means automate it.

06Build-vs-buy worksheet

Six questions to answer in writing before deciding.

From chapter 14. There is no universally right answer - but a written answer to each question makes the right answer for YOUR agency fairly obvious.

01How many senior hours per week does reporting consume today - measured, not guessed?
02Is anyone on the team already comfortable with SQL and a warehouse, honestly?
03What happens to the build if that person leaves - who else can run it?
04Is the next 90 days a stable window, or does client load spike mid-install?
05What does a stalled half-install cost you versus a finished one a month later?
06If you buy: does the vendor hand over documentation your team can own afterward?

Prefer a printable self-assessment? The original Maturity Scorecard PDF is free too.

Want to go further than the book?

Everything above is available worldwide. MB Data Automation's service, training, assessment, product and implementation pages remain limited to U.S. visitors while the reader hub stays globally accessible.

U.S. site: free assessment →U.S. site: free training

The install month, as a checklist.

The chapter 13 sequence - decide, plumb, model, present - broken into checkable steps. Print it, or copy it into your project tool.

Week 1

Decide

Write the KPI dictionary: one line per metric - definition, formula, source
Agree a single UTM naming convention and write it down
Choose the stack (connectors, warehouse, dashboard tool) and record why
Pick the two clients you will install first
Name one owner for the install month

Week 2

Plumb

Connect each source so raw data lands on a schedule - no hand exports
Verify data arrives for 3 consecutive days before building anything on it
Record source quirks you hit (timezones, attribution windows, delays)
Leave raw data raw - transformations belong to next week

Week 3

Model

Encode each KPI dictionary line as a modeled definition over the raw data
Reconcile modeled numbers against the platforms and document known gaps
Add a basic data-quality check on the numbers clients notice most
Have a second person verify one metric end to end

Week 4

Present

Build one templated dashboard reading only the modeled layer
Write the one-page report: headline, three numbers, commentary, next step
Onboard the remaining pilot client from the template - time it
Write the weekly reporting SOP while the steps are fresh

The four stages, with their tell-tale signals.

Chapter 2 in reference form. Read the signals honestly - the stage you are at is the one whose signals make you wince.

Stage 01

Manual

→Reports take 3-6 hours to produce each week
→Every client deck is formatted differently
→Data lives in ad hoc spreadsheets with no version control
→One team member is the single point of failure

Stage 02

Fragile

→Templates exist but require manual cleanup each week
→UTM naming is inconsistent across team members
→GA4 and ad platform numbers frequently disagree
→New clients take 2-3 days to onboard into reporting

Stage 03

Scalable

→Dashboard templates are reused across all client accounts
→KPI definitions are documented and team-wide
→A new account is live in reporting within 2-4 hours
→Reporting SOP exists and is followed consistently

Stage 04

AI-ready

→Data flows automatically: GA4 → BigQuery → dashboard
→AI prompts generate accurate insight commentary on clean data
→Weekly report generation takes under 15 minutes
→Data quality is monitored and alerts fire before clients notice

Eight metrics, defined once.

The starting rows for your own KPI dictionary (chapter 6). Copy them, then adapt the definitions to how YOUR agency counts - the value is that they are written down.

Metric	Definition	Common pitfall
Spend	Total advertising cost for the period, per platform, in account currency.	Currency mixing and partial-day data on the current day.
Impressions	Times an ad was served, as reported by each platform.	Not comparable across platforms - do not sum into one number without a note.
Clicks	Platform-reported ad clicks (not sessions).	Clicks and sessions will never match; decide which one the report leads with.
Sessions	Analytics-reported visits (GA4 definition).	GA4 sessions differ from old UA sessions and from platform clicks.
Conversions	The ONE action you count, named explicitly (e.g. purchase, qualified lead).	The most common source of client mistrust: each platform counts differently. Define it once, in writing.
Revenue	Transaction value attributed to the period, from the source of truth (store/CRM, not the ad platform).	Platform-reported revenue includes modeled attribution; store revenue is what the client reconciles against.
CPA	Spend ÷ conversions, using the definitions above.	Only meaningful once "conversions" is pinned down.
ROAS	Revenue ÷ spend, stating which revenue source is used.	Platform ROAS and blended ROAS are different metrics - label them.

The warehouse-first shape, at a glance.

Sources

Ingestion

Warehouse

Modeled data

Dashboards

AI insight

The Friday routine, in six lines.

01Data check first: did every source load since the last report? If not, fix before writing.

02Scan the numbers clients notice most (spend, conversions, revenue) for anomalies.

03Write the 90-second version first: headline, three numbers, one recommended action.

04Commentary answers "so what" - what changed, why, what we are doing about it.

05Send on the agreed day, every week, even when the news is bad. Consistency is the trust.

06Log anything you fixed by hand - two hand-fixes of the same thing means automate it.

Six questions to answer in writing before deciding.

From chapter 14. There is no universally right answer - but a written answer to each question makes the right answer for YOUR agency fairly obvious.

01How many senior hours per week does reporting consume today - measured, not guessed?

02Is anyone on the team already comfortable with SQL and a warehouse, honestly?

03What happens to the build if that person leaves - who else can run it?

04Is the next 90 days a stable window, or does client load spike mid-install?

05What does a stalled half-install cost you versus a finished one a month later?

06If you buy: does the vendor hand over documentation your team can own afterward?

Prefer a printable self-assessment? The original Maturity Scorecard PDF is free too.