The different steps of the survey ensure that the process meets the requirements for statistical production: 
- tests carried out prior to collection, with 60 companies
- stratified sample selection 
- reminders and checks of data being collected by collectors
- clearance 
- adjustment of total non-response 
- shimming 
- adjustment for partial non-response 
- winsorisation

Before any exploitation, it is useful to assess the quality of the variables through their partial non-response rate and the confidence intervals provided in the quality indicators of the survey.  

Variables with high partial non-response rates should be interpreted with caution, since a non-negligible part of their responses has been imputed.   These include the following variables (from 5% to 31% partial non-response, with variables listed by decreasing partial non-response rate) :     

EXPDESIGN, EXPDPI, EXPSFTW, EXEXPPCT, RMACX, RSERVX, REMPX, EXPMKT, EXPTRAIN, ROTRX, TURNIN, TURNMAR, EXPMAC, TURNNEW, RRDINX, EXAEXP, RRDEXX, and some variables on the type of cooperation partners for innovation activities.