Six heifers were individually housed and assigned to once (FF1) or twice (FF2) daily feeding regimes over backgrounding and finishing periods. Following adaptation, total faecal collections were conducted at 4-h intervals and at 24-h intervals over 4 d, and near-infrared spectroscopy (NIRS) was used to predict faecal organic matter (OM), starch, nitrogen (N), neutral detergent fibre (NDF), acid detergent fibre (ADF), and acid detergent lignin (ADL). At each interval, NIRS calibrations were used to estimate faecal constituents and ADL to calculate apparent (aTTD) and estimated (eTTD) total tract digestibility. Faecal dry matter (DM) (%), NDF, and ADF varied among 4-h interval samples in the backgrounding period and faecal DM, starch, NDF, ADF, and ADL in the finishing period. Faecal starch was able to predict aTTD during both feeding periods (backgrounding: R2 = 0.96, P < 0.01; finishing: R2 = 0.98, P < 0.01). The NIRS calibrations for predicting aTTD using the 4-h interval samples or the 4-d–24-h composite were least accurate for NDF and ADF. Most 4-h interval samples could be used to predict eTTD of nutrients, and aside from starch in the finishing period, there were no differences in eTTD using faecal samples collected over 4-h intervals versus those collected over 4 d. Spot faecal samples collected at any time point from multiple cattle have the potential to predict digestibility. Timing of sampling after feeding must be standardized to predict starch digestibility during the finishing period, with samples between 0–4 h and 8–16 h generating estimates of both starch concentration and digestibility that were closest to that derived from 4-d–24-h composite samples.