Why this matters
A report that takes days to assemble is a symptom, and patching it by hand guarantees the same patching next month. The marketing data quality guide gives the failure catalogue and the weekly routine; the guide to data operations and taxonomies gives the taxonomy and controls at the point of creation. At scale, failures cross platforms, currencies and teams, so most fixes need an owner outside analytics.
The operator and its monthly scramble
Coachlark Tours is an invented coach tour operator selling escorted holidays to customers in the UK, Ireland, the Netherlands and Germany. Each market has its own ad accounts; one analytics property (GA4Google's event-based analytics platform.) runs in UK time and pounds; bookings are recorded in a booking system with booking date, departure date and sales market.
Coachlark Tours: one month's campaign extract
| Row | Platform and account | Currency | Campaign name | Source / medium / campaign on links | Spend | Platform bookings |
|---|---|---|---|---|---|---|
| A | Search, UK | GBP | UK_Brand_Search | Auto-tagged | 4,200 | 310 |
| B | Search, Germany | EUR | Brand-DE-Search | Auto-tagged | 2,900 | 95 |
| C | Search, Netherlands | EUR | nl brand search | Auto-tagged | 1,100 | 40 |
| D | Social, UK | GBP | Rhine Spring 26 | facebook / paid_social / rhine-spring | 3,000 | 120 |
| E | Social, Europe | EUR | rhine spring DE | Facebook / Social / RhineSpring | 2,500 | 70 |
| F | Social, Europe | EUR | rhine spring NL | facebook / paid_social / rhine-spring, copied from the UK ad set | 1,200 | 35 |
| G | GBP | March newsletter | Tags on 4 of 9 links | n/a | n/a | |
| H | Affiliate network | GBP | Network default | Affiliate / affiliates | 1,800 | 60 |
What the report showed:
- Total spend 16,700, adding pounds and euros in one column.
- German spend converted at the spot rate on report day this month and at the budget rate last month.
- Row E's sessions in Organic Social; row F's bookings credited to the UK; untagged email visits in Direct; affiliate traffic split across two channel rows.
- Analytics purchases include 38 duplicates with a repeated transaction ID, from confirmation-page reloads.
- Marketing reports bookings by booking date; finance reports revenue by departure date.
Budget rate for the year, agreed with finance: €1 = £0.85.
From failure log to root causes
Log every symptom, then ask where it was created: a hand-typed name, a missing rule, a platform default, a tag, or two valid definitions sharing one word. Fix it where it is born, with an owner who works there, and record every fix in a document so it is not undone.
Task
- Log each failure with its symptom, root cause, fix at source and owner.
- Write the campaign taxonomy and the naming and tagging rules for every platform.
- Set the weekly checks, each with an owner and a tolerance.
- Define how market results are converted, dated and consolidated, and restate this month's spend.
Worked answer
Failure log.
| Failure | Root cause | Fix at source | Owner |
|---|---|---|---|
| Names in several styles (A to C) | No taxonomy; names typed by hand | Taxonomy and a name builder in the intake form | Marketing operations |
| Paid social read as organic (E) | Medium typed outside a controlled list | Builder with fixed values, lower case | Marketing operations |
| Dutch bookings credited to the UK (F) | Ad set copied without updating tags | Launch check comparing live tags with the work item | Social lead |
| Email in Direct (G) | Links tagged by hand in each send | Template that tags every link automatically | Customer relationship manager |
| Affiliate split in two (H) | Network default differs from ours | Set the tracking template once in the network | Affiliate manager |
| Currencies added together | Report built by pasting platform exports | Rate table in the warehouse; convert on load | Analyst |
| Rates change month to month | No written rule | Rule in the measurement plan | Head of analytics, with finance |
| Duplicate purchases | Purchase tag fires on reload | Fire once per transaction ID; deduplicate in the warehouse | Analytics |
| Marketing and finance disagree | Two valid date rules under one name | Two named metrics: bookings by booking date, revenue by departure date | Analyst and finance |
Taxonomy and rules. Attributes: market, objective, product, audience, launch month, variant, each from a list owned by marketing operations with a change log. Campaign names follow market_objective_product_audience_month in lower case, such as de_prospecting_rhine_families plus the launch month. Search keeps auto-tagging on with the same names; social uses the pattern for campaigns, the audience for ad sets and the variant for ads. Every manual UTMURL tags identifying traffic source. link comes from the builder: source is the platform, medium from a fixed list, campaign is the campaign name exactly, content is the variant. The UTM and naming conventions guide sets case and separator rules.
Weekly checks.
| Check | Owner | Tolerance |
|---|---|---|
| New campaign names that do not parse | Marketing operations | None; fix within the week |
| Paid sources appearing in organic or unassigned channels | Analyst | Near zero |
| Analytics purchases against booking system bookings | Analyst | Ratio within the recorded band |
| Rate table and spend by account currency | Analyst | Matches finance's rate file |
| Conversion settings changed in any platform | Channel leads | Every change logged |
Consolidation. Report in pounds at the budget rate for performance against plan, local currency alongside, and reconcile with finance at actual rates separately. Take the market from the booking system, never from visitor location. Date bookings by booking date in UK time for marketing, and revenue by departure date for finance. Keep shared definitions, market targets and data notes, as the guide to measuring and reporting across markets recommends, so each headline KPIA metric tied directly to an objective. compares like with like.
This month's spend restated
Euro accounts: €2,900 + €1,100 + €2,500 + €1,200 = €7,700 × 0.85 = £6,545. Pound accounts: £4,200 + £3,000 + £1,800 = £9,000. Total: £15,545, not the 16,700 the report showed, which added the two currencies as if they were one. German spend alone, €2,900 at the budget rate, is £2,465 in the performance view; finance's reconciliation at its actual monthly rate sits on a separate line, so a currency movement never reads as a change in cost per booking.
Common mistakes
- Cleaning the report, not the source. Next month needs the same cleaning.
- A taxonomy nobody owns. New values appear and the parsing check fails silently.
- Converting at whatever rate is handy. Currency movement then looks like performance.
Bring this governance into the analytics and reporting plan capstone, or revisit metric definitions at the intermediate marketing analytics level.
