Retail measurement service providers maintain global
standards for sampling error, which define the acceptable tolerance level of error at a
specified level of confidence for their retail audits. One example is NielsenIQ, which has
established the following global standard for sampling error at a 90% level of confidence,
applicable to categories available in 80% of the universe:
- National Market: ± 3% of sales level
- Major Market Breakdowns/Channels: ± 6% of sales level
- Minor Market Breakdowns: ± 6 to 10% of sales level
According to this standard, the sample should be designed in such a way that,
for a national market, there is a 90% probability that the estimated sales value will fall
within ± 3% of the actual value.
Note that the accuracy standards are dependent on the availability of the
category being measured. If the category is only present in a small proportion of stores, the
effective sample size for that category is lower, which adversely affects data accuracy.
Therefore, the global standard is applicable to categories that are available in at least 80%
of the universe of stores.
