Research Methodology

Data sources

Pitch deck research draws from three corpora:

• Public deck galleries: Y Combinator showcase decks, AngelList public profiles, Slidebean public gallery, Pitchdeckcoach examples. We use only decks the founders or accelerators have explicitly published as public references.
• Founder interviews: 30–45 minute structured interviews with founders who have raised in the last 24 months. Quotes are attributed by name when permission is granted; aggregate statistics are anonymized.
• SlideGMM platform aggregates: when used, we report only macro-level counts (slide count distribution, color palette frequency, export-format ratios). No deck content, no user identifiers, no PII ever leaves the database — and we flag explicitly when a stat comes from platform data so you can weight the conflict-of-interest accordingly.

What we exclude

• Decks behind paywalls or non-public NDAs. We don't pay for access; if a deck isn't reachable from a public URL, it doesn't enter the corpus.
• Decks older than 5 years for trend analysis. Pitch deck conventions shift fast — citing a 2017 deck as evidence for what works in 2026 is misleading.
• Decks where the founder's outcome is unknown. We track fundraising results when public so we can correlate deck attributes with success — decks with no outcome data are still counted in structural analysis but excluded from outcome-correlation claims.

Conflicts of interest

We make a presentation product. That means our research is inherently interested — we want SlideGMM to win. We mitigate this by:

• Naming our product's weaknesses explicitly in comparison posts. If a competitor wins a category, we say so.
• Publishing raw datasets when feasible (CSV / JSON downloads alongside the report) so independent analysts can re-run the numbers.
• Disclosing platform-data origin on every stat that uses it. Stats from external corpora are weighted higher than platform-internal stats in headline claims.

Comparison-post methodology

Comparison pages (/compare/slidegmm-vs-…) follow this fixed evaluation structure:

1. Pricing: verified from the competitor's public pricing page within the last quarter. We capture the date of last verification at the top of each comparison.
2. Feature parity: tested on a real account on both platforms. We do not infer features from marketing pages.
3. Export quality: exported the same source deck from both tools and opened the .pptx in Microsoft PowerPoint. Visual fidelity, font preservation, and edit-ability scored.
4. "When to choose [competitor] instead" section is required on every comparison post. If we can't identify a use case where the competitor wins, the comparison isn't honest enough to publish.

Update cadence

Comparison posts are re-verified at least quarterly. The dateModified field at the top of each post reflects the last verification date. Pricing changes or feature launches from a competitor trigger an immediate update.

Research reports are static — we don't edit numbers retroactively. If a number turns out wrong, we add a correction entry to /about/corrections and link the correction from the original report. The original number stays visible (struck through) so the audit trail is preserved.

Questions or corrections

If a stat looks wrong, our methodology has a hole, or you can point at primary data we missed: email research@slidegmm.ai. We'll respond within a week and add anything we can verify to the corrections log.