À propos de cet exercice
Un jeu de données de 300 lignes sur une feuille, 3 cellules de réponse, environ 8 minutes. Corrigé dans votre navigateur : la valeur et la formule qui la produit, avec des indications sur la correction à apporter.
Comment le résoudre
- G2
=UNIQUE(TRIM(C2:C301)) - J2
=ROWS(UNIQUE(TRIM(C2:C301))) - F2
=COUNTIF($C$2:$C$301,TRIM(C2)&"*")
UNIQUE returns every distinct value of a range once and spills the result into the cells below, so one formula in G2 produces the whole customer list, in the order the names first appear. It ignores letter case — "oakridge marine" and "Oakridge Marine" are one customer — but not spaces, which is why TRIM inside the range is what folds "Ridgeway Analytics " into "Ridgeway Analytics": 58 customers instead of 70. ROWS around the same UNIQUE turns the list into a count, and COUNTIF with the pattern TRIM(C2)&"*" answers the per-order question: how many rows belong to this customer, stray spaces included.
Pitfall 1: Remove Duplicates (Data tab) is destructive — it deletes rows, keeps the first occurrence and has no live link to the source; UNIQUE is a formula, so a new order appears in the list the moment it is typed. Both ignore letter case; neither trims spaces. Pitfall 2: #SPILL! appears when anything sits in the cells the array needs (here G3:G59); clear the path or move the anchor. Pitfall 3: COUNTIF compares like Remove Duplicates — case-insensitive, space-sensitive — and needs a range as its first argument, so it cannot TRIM the column: =COUNTIF($C$2:$C$301,C2) misses the trailing-space copies and is wrong on 78 rows here. The wildcard pattern TRIM(C2)&"*" fixes that, but it would over-count if one customer's name were the start of another's ("Summit Labs" and "Summit Labs Group" — none here); the exact version is =SUMPRODUCT(--(TRIM($C$2:$C$301)=TRIM(C2))), which trims the whole column in memory and is slower on long lists. Alternatives: =SORT(UNIQUE(TRIM(C2:C301))) for an alphabetical list; =UNIQUE(TRIM(C2:C301),,TRUE) lists only the 8 one-off customers; the pre-365 distinct count =SUMPRODUCT(1/COUNTIF(C2:C301,C2:C301)) cannot trim and returns 70. Version note: UNIQUE, SORT and spilling need Excel 365 or Excel 2021; in Excel 2019 put =TRIM(C2) in a helper column, paste it as values and run Data → Remove Duplicates on a copy. Desktop Excel can refer to the whole spill as G2# (=COUNTA(G2#)); this in-browser grid does not support # yet.
Autres méthodes valables
- =COUNTA(G2:G301) — counts the spilled list for J2 and grows with it
- =SUMPRODUCT(--(TRIM($C$2:$C$301)=TRIM(C2))) — the exact per-order count: compares every trimmed name with this one, no wildcard needed
- =SORT(UNIQUE(TRIM(C2:C301))) — the same names in alphabetical order (a valid list, but this check compares the first-appearance order)
- =SUMPRODUCT(1/COUNTIF(C2:C301,C2:C301)) — pre-365 distinct count; returns 70 because it cannot TRIM
Erreurs fréquentes
Writing =UNIQUE(C2:C301) without TRIM: the list looks right at a glance but holds 70 entries instead of 58 — from G16 on, every name is one row off because "Ridgeway Analytics " is listed again.
Comment nous corrigeons
Toute formule qui produit le bon résultat est acceptée, sauf si la tâche nomme une fonction : une autre fonction rapporte alors la moitié des points. Nous comparons les valeurs avec une petite tolérance et ignorons la casse et les espaces dans le texte. Un nombre correct saisi à la main au lieu d’une formule rapporte la moitié des points. Les tâches sur plusieurs cellules rapportent tous les points avec 100 % de cellules justes, la moitié à partir de 50 %. Lorsque c’est pertinent, nous modifions aussi une cellule d’entrée sur une copie de votre feuille et vérifions que votre formule suit.
Réponses calculées selon les règles d’Excel 365 et vérifiées dans notre correcteur le 16 novembre 2026. Auteur : GridBelt team.