The Air Goes Out of the Room: Calls Where Pricing Reads as Under Pressure
We examined 41,241 earnings calls — 25.0% of a 165,182-call corpus spanning 1990 to 2026 — in which AI-assisted reads of the transcript flagged pricing as under pressure. On the language profile, these calls show lower confidence (6.76 vs. 7.21) and promotion (4.61 vs. 5.05), and higher stress (2.90 vs. 2.43). Guidance behavior diverges too: 19.5% lowered guidance versus 11.6% in the base set, and 12.2% raised it versus 21.1%. Where forward returns were measurable (5,222 calls), the median was -8.7% versus -7.2% in the base, with 37.4% beating expectations against 39.5% overall. The paper describes these patterns without asserting any causal link or trading edge.
- Pricing-pressure calls show confidence of 6.76 versus 7.21 in the broader corpus, and promotion of 4.61 versus 5.05.
- 19.5% of these calls lowered guidance, compared with 11.6% of the base set, while 12.2% raised guidance against 21.1%.
- The 2015-2025 trend falls from 40.4% of calls to 21.0%, with a low of 15.6% in 2021.
- Among 5,222 calls with measurable forward returns, the median was -8.7% versus -7.2% for the base, and 37.4% beat expectations versus 39.5%.
- The strongest over-represented tag is 'Results Worse Than Direction' at 1.25x lift (64.0% vs. 51.1%), while 'Deferred Revenue Growing' appears at 0.49x.
1Introduction
Pricing is the lever management teams least like to discuss and analysts most like to probe. When a call's language suggests pricing is under pressure, the transcript often carries the tension in indirect ways: hedged answers, softer promotion, tightened guidance. For anyone who follows earnings calls closely, these markers matter because pricing power is central to how margins and growth narratives are judged — and because transcript tone is one of the few things observable in real time. This study examines 41,241 calls flagged as pricing-under-pressure, describing their language profile, guidance behavior, recurring tags, frequency over 2015-2025, and forward returns, purely as a descriptive exercise.
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 pricing was read as under pressure (n = 41,241; 25.0% of the reference set, 95% Wilson interval 24.8%–25.2%). 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
The language profile is the clearest signal: confidence sits at 6.76 versus 7.21, promotion at 4.61 versus 5.05, and stress at 2.90 versus 2.43, while candor (7.02 vs. 6.86) and specificity (7.58 vs. 7.56) barely move — these calls are not less forthcoming, just less upbeat. Guidance skews defensive: 19.5% lowered versus 11.6%, and 3.7% withdrew versus 2.7%. Tag lifts reinforce the picture — 'Results Worse Than Direction' appears at 1.25x (64.0% vs. 51.1%), while 'Deferred Revenue Growing' (0.49x) and 'A Tiny Fraction of the Market' (0.71x) are under-represented. The share of such calls fell from 40.4% in 2015 to 21.0% in 2025, bottoming at 15.6% in 2021. Median forward returns were -8.7% versus -7.2%.
| Meter | Study group | Baseline | Δ |
|---|---|---|---|
| Candor | 7.02 | 6.86 | +0.16 |
| Evasion | 2.79 | 2.70 | +0.09 |
| Specificity | 7.58 | 7.56 | +0.02 |
| Stress | 2.90 | 2.43 | +0.47 |
| Promotion | 4.61 | 5.05 | -0.44 |
| Confidence | 6.76 | 7.21 | -0.45 |
| Action | Study group | Baseline |
|---|---|---|
| Raised | 12.2% | 21.1% |
| Maintained | 50.1% | 48.8% |
| Lowered | 19.5% | 11.6% |
| Withdrawn | 3.7% | 2.7% |
| Signal | Lift | In group | Baseline |
|---|---|---|---|
| Results Worse Than Direction | 1.25× | 64.0% | 51.1% |
| Deferred Revenue Growing | 0.49× | 4.3% | 8.9% |
| A Tiny Fraction of the Market | 0.71× | 21.2% | 30.0% |
| Statistic | Study group | Returns sample |
|---|---|---|
| Median excess return | -8.7% | -7.2% |
| Interquartile range | -27.3% to +10.2% | — |
| Share beating SPY | 37.4% (95% CI 36%–39%) | 39.5% |
| Observations | 5,222 | 22,449 |
| Ticker | Quarter | Call date | Call grade |
|---|---|---|---|
| AON | Q2 2025 | 2025-07-25 | C |
| NWG | Q2 2025 | 2025-07-25 | B+ |
| AMSF | Q2 2025 | 2025-07-25 | C+ |
| VRTS | Q2 2025 | 2025-07-25 | C+ |
| KNSL | Q2 2025 | 2025-07-25 | C+ |
| HTH | Q2 2025 | 2025-07-25 | B |
| STEL | Q2 2025 | 2025-07-25 | B |
| FISI | Q2 2025 | 2025-07-25 | B+ |
4Discussion
A careful reader should treat this as a description of how certain calls sound and what accompanies them, not as a rule. Calls flagged for pricing pressure tend to be less promotional, more stressed, and more likely to cut guidance — but none of these co-occurrences tells you which direction, if any, drives the others, and the returns differences are modest in context. The declining 2015-2025 trend may reflect genuine conditions, changing disclosure habits, or how the detector behaves across eras. Nothing here supports predicting outcomes or acting on any single call's tone.
5Limitations
All transcript fields here are AI-generated reads and inherently noisy; a 'pricing pressure' flag is an interpretation, not a fact about the business. The returns sample covers 5,222 flagged calls out of a base of 22,449, skewed toward liquid names, so the return comparison may not generalize. Our own forward tests falsified directional prediction from these signals. Finally, LLMs partially remember famous stocks' histories, which contaminates any backtest of transcript-derived features. See the full methodology, including the C1 pattern’s forward-test failure and the LLM-memorization finding.