We have produced original research reports for clients that got cited by journalists for two years afterward, and others that were read once, by nobody outside the company that commissioned them. The difference was almost never the sample size or the survey design. It was whether the finding was surprising to the audience it was written for.
The pattern in reports that get cited repeatedly
Every report of ours that got sustained pickup contained at least one number that contradicted a widely-held assumption in its industry. "Most companies think X, but our data shows Y" is a more citable sentence than any amount of confirmatory data, because it gives a journalist or a fellow marketer a reason to reference it beyond politeness.
The mistake we see most often
Commissioning a survey to confirm something the company already believes, then publishing the confirmation as if it were news. Confirmatory findings are useful internally for a sales deck. They are almost never a marketing asset on their own.
What we changed in our own process
We now write the hypothesis we are hoping to disprove before we write a single survey question, and we design the questions specifically to be able to surface a contrarian finding if one exists in the data, rather than only measuring what we expect to find.
Aim for one genuinely surprising, specific number per report, not ten mildly interesting ones.
Publish the methodology in full — citability depends on other people being able to trust the number.
Re-run the same survey annually if the first finding lands; year-over-year change is its own story.
