Relative size of error bars of ΛCDM parameters (Ω_ch², Ω_bh²,H₀, n_s, log A_s) for different bin sizes Δℓ with respect to Δℓ = 50 for a mock CMB data set. In the rightmost panel, we show the determinant of the parameter covariance 𝒞, a proxy for the five-dimensional volume, using a different scale for the vertical axis. We note that the parameter that is affected the most by the broadening of bins is H₀, which relies on an accurate measurement of the acoustic peak positions. Due to the quadratic relationship between Δℓ and σ, bin sizes smaller than Δℓ = 50 yield a negligible improvement on marginalised parameter errors (≤3%). Forecasts are fast and easy with candl, enabling thorough analysis optimisation.