Claim
The two content formats that resist AI commoditization are original research (data that does not yet exist on the internet) and opinionated thought leadership. Of the two, original research is the most under-used and most defensible because it generates new statistics that journalists and other publishers must cite. One annual study can out-link a full year of weekly publishing.
Mechanism
Most content describes existing knowledge. Original research creates knowledge — a stat, a benchmark, a survey result that did not exist before publication. Citation behavior is asymmetric: anyone writing about the topic in the next decade may link to the source of the stat, compounding backlinks and authority. Crestodina's annual Blogger Survey, run for 10+ years, has accumulated 2,500+ backlinks to a single URL.
Conditions
Holds when:
- The team has access to a community or panel from which to collect data.
- The study question is genuinely under-studied (not a republishing of existing benchmarks).
Fails when:
- The audience is too small to generate statistically defensible sample sizes.
- The "research" is thinly disguised marketing — readers and journalists detect and ignore.
Evidence
"He has practiced this himself for over a decade with the annual Blogger Survey, which has attracted 2,500+ backlinks to a single URL."
"Publish fewer pieces, but invest disproportionately in two formats that resist AI commoditization. The first is original research..."
— Andy Crestodina (synthesized from operator's published work)
Signals
- Editorial calendar reserves a "research project" slot once or twice a year, with a separate budget.
- The team owns a survey panel, customer base, or community used as the data source.
- Marketing reports track lifetime backlinks per asset, not just first-90-day traffic.
Counter-evidence
For very small teams or early-stage startups, the upfront cost of designing, running, and writing a study can starve the cadence content that builds initial audience. A "build-the-research-engine-first" sequence may be premature for teams without baseline reach.
Cross-references
- (none in current corpus)