Three Call Centres: A Reporting Change Disguised as Improvement

Restate South first, because the asterisk decides the headline.

• Q2 answered calls = 24,700 − 1,606 = 23,094.
• 66% of answered = 15,242 calls answered within 20 s.
• As a percentage of calls offered: 15,242 ÷ 24,700 = 61.7%.

So South moved from 64.0% to 61.7% - a fall of 2.3 points, not a 2-point rise. The 66% is not a better number, it is a different fraction: changing the denominator from offered to answered removes abandoned calls from the base entirely. That flatters the ratio precisely when abandonment is worsening, and here it is - South's abandonment rate rose from 6.0% (1,185/19,750) to 6.5% (1,606/24,700).

Like-for-like movement:

• North: 71 → 68, down 3 points on the largest base.
• Central: 78 → 81, up 3 points.
• South: 64 → 61.7, down 2.3 points.

Central is the only region that improved. The director's statement is the reverse of the evidence: South is the only region whose published figure improved while its underlying performance got worse.

Q1 (89,550 calls offered):

• North 0.71 × 41,200 = 29,252
• Central 0.78 × 28,600 = 22,308
• South 0.64 × 19,750 = 12,640
• Total 64,200 ÷ 89,550 = 71.7%

Q2, like-for-like (97,100 calls offered):

• North 0.68 × 45,100 = 30,668
• Central 0.81 × 27,300 = 22,113
• South 0.617 × 24,700 ≈ 15,242
• Total 68,023 ÷ 97,100 = 70.0%

Group movement: down 1.7 points, from 71.7% to 70.0%.

Decomposition in calls answered within 20 s:

• North: 45,100 × (0.68 − 0.71) = −1,353
• Central: 27,300 × 0.03 = +819
• South: 24,700 × (0.617 − 0.64) ≈ −568
• Net −1,102 calls; −1,102 ÷ 97,100 ≈ −1.1 points of rate effect.

Mix effect: applying Q1 rates to Q2 volumes gives 69,123 ÷ 97,100 = 71.2%, so shifting volume toward the weaker regions costs about a further −0.5 point. The two effects roughly reconcile to the 1.7-point fall.

Why the group fell while a region rose: North is the largest base. A 3-point slide on 45,100 calls destroys more good answers than Central's 3-point gain on 27,300 creates, and Central is the only contributor on the positive side. Group figures are volume-weighted; a single regional improvement can be swamped.

The flaw is that the premise is an artefact of the measurement, not a result.

• On a like-for-like basis South went backwards (64.0% → 61.7%) and its abandonment rate rose. There is no measured improvement to roll out.
• Even the raw denominators are confounded: South's offered volume grew 25% (19,750 → 24,700) while North grew 9.5% and Central shrank 4.5%. A queue redesign implemented in the same quarter as a volume surge cannot be separated from it on this data. Staffing, call mix and average handling time are all absent.
• The 66% is internally useful to South's manager - answered-call service level is a legitimate measure of what happens once a caller gets through - but presenting it next to the others without a footnote converts a definitional choice into an apparent trend.

The one check to insist on: confirm South's Q2 definition matches the other regions on all three parameters - denominator (offered vs answered), the 20-second threshold, and the treatment of abandoned and short-abandon calls - then restate Q1 under it. If Q1 and Q2 cannot be reconciled on a single definition, there is no measured change at all and the rollout has no evidence base.

What I would ask for before any rollout decision: South's Q2 staffing hours against Q1, its average handling time, and its repeat-call rate. A 'design' that improves speed by shrinking talk time or deflecting calls is not the same improvement as one that answers more calls within 20 seconds.