In this study, predictions obtained in the four and in the five
flavour schemes are compared for two important processes involving heavy
flavours at the LHC: the production of a $Z$ or a Higgs boson in association
with $b$ quarks. In particular we obtain predictions with \Sherpa's
\MCatNLO implementation for the four--flavour scheme, treating the $b$'s
as massive, and with multijet merging at leading and next-to leading
order for the five--flavour scheme.

While differences between the two schemes, at the inclusive level,
are well understood from resummation of possibly large logs into
the $b$-PDFs, differences in shape present a major problem
for experimental measurements. We make use of data for $Z+b(\bar{b})$
production at the $7$ TeV LHC to exhibit strengths and weaknesses of the
different approaches and we use these results to validate predictions for
$b$-associated Higgs-boson production at the 13 TeV Run II.