Build It and They Will Say It's Growing: 'Early Products Growing Fast' on Earnings Calls
This study examines 63,629 earnings calls from 1990 to 2026 in which an AI model answered YES to the battery item "Early Products Growing Fast" — 38.5% of the 165,182-call corpus (95% CI 38.3%–38.8%). These calls skew promotional: promotion scores 5.60 vs. 5.05 elsewhere and confidence 7.44 vs. 7.21, while candor runs slightly lower (6.72 vs. 6.86). Management raised guidance on 27.2% of these calls versus 21.1% of others. Language lifts cluster around growth narratives: "A Tiny Fraction of the Market" (1.84x), "Deferred Revenue Growing" (1.67x), and "The Hidden Segment" (1.55x). Among 9,514 calls with forward returns, the median was -6.5% and 41.7% beat the median of the broader 22,449-call sample. The share of such calls peaked at 46.8% in 2021.
- 38.5% of 165,182 calls (n=63,629) were flagged YES for "Early Products Growing Fast," with a 95% CI of 38.3% to 38.8%.
- These calls score higher on promotion (5.60 vs. 5.05) and confidence (7.44 vs. 7.21) but slightly lower on candor (6.72 vs. 6.86).
- Guidance was raised on 27.2% of flagged calls versus 21.1% of unflagged calls, and the phrase "A Tiny Fraction of the Market" appears 1.84x more often.
- Among 9,514 flagged calls with forward returns, the median return was -6.5%, and 41.7% beat the sample median versus 39.5% in the 22,449-call base.
1Introduction
Few claims on an earnings call are as seductive as a young product line growing fast. It reframes the whole conversation: today's numbers matter less than tomorrow's curve. For anyone who dissects calls for a living, knowing how often this framing appears — and what accompanies it — is useful context. Is the growth story paired with candor and specificity, or with promotion and confidence? Does it travel with raised guidance, and with the classic vocabulary of the hidden-gem pitch? This study examines 63,629 calls from 1990 to 2026 where a model answered YES to the battery item "Early Products Growing Fast," describing their language profile, guidance behavior, phrase lifts, prevalence over time, and subsequent returns.
2Data & methodology
The corpus comprises 165,182 earnings-call transcripts published between 1990 and 2026, each scored independently by a large language model on an identical 37-field battery: seven categorical business verdicts, eight 0–9 behavioral meters, and twenty yes/no judgments. The study group is defined as calls where the model answered YES to the battery item "Early Products Growing Fast" (n = 63,629; 38.5% of the reference set, 95% Wilson interval 38.3%–38.8%). Baseline figures use all scored calls. Market outcomes join a fixed sample of 22,449 calls with twelve-month total returns in excess of SPY, measured from the first close after each call; this sample skews toward liquid U.S. names and is reported as descriptive history only.
3Results
Flagged calls carry a promotional accent: promotion averages 5.60 versus 5.05 on other calls, confidence 7.44 versus 7.21, while candor is slightly lower at 6.72 versus 6.86. Guidance actions lean optimistic — raised on 27.2% of flagged calls versus 21.1% otherwise, maintained on 46.7% versus 48.8%. The phrase lifts read like a growth-pitch anthology: "A Tiny Fraction of the Market" at 1.84x, "Deferred Revenue Growing" at 1.67x, "The Hidden Segment" at 1.55x, and "Volume About to Step Up" at 1.45x. Prevalence climbed from 33.0% in 2015 to a 46.8% peak in 2021 before settling near 39.2% in 2024. Forward returns were unremarkable: a median of -6.5% across 9,514 calls, with 41.7% beating the sample median versus 39.5% in the base.
| Meter | Study group | Baseline | Δ |
|---|---|---|---|
| Candor | 6.72 | 6.86 | -0.14 |
| Evasion | 2.79 | 2.70 | +0.09 |
| Specificity | 7.54 | 7.56 | -0.02 |
| Stress | 2.38 | 2.43 | -0.05 |
| Promotion | 5.60 | 5.05 | +0.55 |
| Confidence | 7.44 | 7.21 | +0.23 |
| Action | Study group | Baseline |
|---|---|---|
| Raised | 27.2% | 21.1% |
| Maintained | 46.7% | 48.8% |
| Lowered | 10.6% | 11.6% |
| Withdrawn | 2.2% | 2.7% |
| Signal | Lift | In group | Baseline |
|---|---|---|---|
| A Tiny Fraction of the Market | 1.84× | 55.2% | 30.0% |
| Deferred Revenue Growing | 1.67× | 14.8% | 8.9% |
| The Hidden Segment | 1.55× | 32.7% | 21.1% |
| Volume About to Step Up | 1.45× | 41.2% | 28.5% |
| Founder-Led Companies | 1.38× | 27.6% | 20.0% |
| Statistic | Study group | Returns sample |
|---|---|---|
| Median excess return | -6.5% | -7.2% |
| Interquartile range | -28.3% to +15.2% | — |
| Share beating SPY | 41.7% (95% CI 41%–43%) | 39.5% |
| Observations | 9,514 | 22,449 |
| Ticker | Quarter | Call date | Call grade |
|---|---|---|---|
| AON | Q2 2025 | 2025-07-25 | C |
| OMF | Q2 2025 | 2025-07-25 | A |
| VRTS | Q2 2025 | 2025-07-25 | C+ |
| FLG | Q2 2025 | 2025-07-25 | B |
| FRST | Q2 2025 | 2025-07-25 | A |
| AN | Q2 2025 | 2025-07-25 | B |
| CHTR | Q2 2025 | 2025-07-25 | C+ |
| DBOEY | Q2 2025 | 2025-07-25 | B+ |
4Discussion
A careful reader should conclude that the "early products growing fast" framing is common, slightly promotional in tone, and associated in this sample with more frequent guidance raises and the vocabulary of emerging-opportunity pitches. One should not conclude that the framing causes better outcomes, that these companies outperform, or that the modest differences in beat rates signal an edge — the differences are small and the study is descriptive, not predictive. The 2021 peak coincides with a period of broad growth optimism, but prevalence trends reflect language and labeling, not verified fundamentals.
5Limitations
The battery items are AI-read fields and inherently noisy; a YES label reflects model judgment of call language, not audited fact. The returns sample covers 22,449 calls, of which 9,514 are flagged, and is skewed toward liquid names, limiting generalizability. Our own forward tests falsified directional prediction from these features, so nothing here should be read as a trading signal. Finally, LLMs partially remember famous stocks' histories, contaminating any backtest built on model-labeled language. See the full methodology, including the C1 pattern’s forward-test failure and the LLM-memorization finding.