ILG Ads · Convention v2 · Effective 1 Oct 2026
Campaign naming convention
One 3-letter division code at the front of every campaign name. The monthly report reads the code instead of guessing from keywords.
NUR
-
LeadGen-5kmRadius-Pmax-16thOct2025-Collingwood-IOS
Division code
The only thing you add. Exactly one, always first.
Existing name — unchanged
Keep radius, date, format and school name exactly as they are. Remove any now-redundant division word.
Front, always
Meta truncates long names; a leading code stays visible, sorts campaigns by division, and can't be mistaken for a date.
Three letters, not words
PRP can't hide inside another word. PREP sits inside PrepSchool — that's the ambiguity we're removing.
No school codes
80 of 81 flagship ad sets already resolve from the school name. Just spell each school the same way every time.
NUR
Nursery
Nursery, kindergarten, montessori, pre-school, Stay & Play.
PRP
Prep
Prep school, Reception, junior years.
SEN
Senior & Sixth Form
Senior school, Sixth Form, 6th Form.
FUL
Full School
Whole-school or brand campaigns, and anything covering more than one division.
EXC
Excluded
Swim School, camps, activity clubs — anything that isn't admissions.
Exactly one code per campaign. NUR-PRP- is invalid — a multi-division campaign is FUL. Typos like SNR or PREP silently fall back to guessing.
Is it an admissions campaign?
No
EXC
Does it cover more than one division, or the whole school?
Yes
FUL
One division only — which one?
NUR
Nursery · kindergarten · montessori · pre-school · Stay & Play
PRP
Prep school · Reception · junior years
SEN
Senior school · Sixth Form · 6th Form
4 · Where it gets applied
Google Ads · 32 accounts
Campaign nameRequired
Ad set namen/a
Google reports pull at campaign level.
Meta · ILG flagship
Campaign nameRequired
Ad set nameRequired
Meta reports pull at ad set level — a code on only one level won't cover both.
Meta · standalone
Campaign nameRequired
Ad set nameRequired
St Felix · Rookwood · Derby Grammar · Shebbear · St Francis Pewsey · Yateley Manor
{{ row.before }}
→
{{ row.code }}{{ row.after }}
6 · How the report reads it
Campaign or ad set name
Does it start with a code?
Yes
Division read directly from the prefix. Exact, no guessing.
No
Falls back to today's keyword logic, so legacy campaigns keep working during rollout.
Each monthly report shows classified by code vs. by inference — so migration progress is visible.
Rename on the 1st
Never mid-month. A month split across two conventions is harder to report on than either alone.
Keep the mapping sheet
Old name → new name, every campaign. The one step that can't be redone later.
Learning is safe
Both platforms track the entity ID, not the label. Performance and learning phase are unaffected.
7 · What it unlocks for AI reporting
Every AI-generated report starts by classifying rows. Today that classification is inference — pattern-matching on words that may or may not be there. A prefix turns it into a lookup, and everything downstream inherits the certainty.
Today — inference
Keyword rules and hand-maintained spelling variants (Westbury / WestburyHouse) sit in the reporting code.
A new campaign name can silently misclassify — the report still looks correct.
Every new question ("split Prep by school") means re-checking the guesswork first.
With codes — lookup
Division is a field, not a judgement. The same name always produces the same row.
Misclassification becomes visible: a missing prefix is a flagged gap, not a silent default.
New cuts of the data are just a filter — no reporting-code change, no re-verification.
Trustworthy CPL
Camps and Swim School can't leak into admissions spend, so cost per lead is comparable school to school and month to month.
Cleaner trends
Divisions are defined the same way every month, so year-on-year and month-on-month movements reflect performance, not naming drift.
Answers on request
"Nursery spend across all 32 accounts, last quarter" becomes a direct query — no per-school interpretation step in between.
Auditable output
Each report states how many rows came from a code and how many from inference — the confidence in the numbers is on the page.