Real-Time Survey Triangulation
Pair consistent company surveys with lagged official data to see the economy now
- Difficulty
- Advanced
- Time to result
- ~ongoing to results
- Steps
- 6
- Confidence
- 98%
Hyman's method builds a current economic picture from repeated company surveys and then checks it against official statistics. The survey asks the same person the same question each period: how current sales compare with expectations. It spans many industries, preserves continuity when a respondent is temporarily absent, and gives participating companies the aggregate result and economic research in return. Because revenue combines activity and price, the analyst examines what drives a move rather than treating every increase as pure volume. These coincident readings arrive before many government releases. The final mechanism is triangulation: compare the surveys with retail sales, GDP, housing starts, employment, and other official series; identify whether one source leads or lags; and revise the present-tense view as evidence changes. The output is a disciplined nowcast, not a claim of perfect prediction.
Origin
Hyman expanded an existing weekly retailer survey into roughly thirty areas because he saw that the same current-activity question could reveal conditions across the economy.
Core principles
- 01Ask about current activity rather than distant expectations.
- 02Keep the respondent and question consistent across observations.
- 03Give participants useful information in return for candid answers.
- 04Treat private surveys and official statistics as complementary evidence.
- 05Investigate leads, lags, and divergences instead of trusting one source.
How to run it
- 1
Map the economy
Choose recurring panels across industries such as retail, trucking, home building, manufacturing, airlines, and overseas-facing businesses. Cover enough distinct activity to avoid mistaking one sector for the whole economy.
Pro tip Include industries that reveal both physical activity and changing prices.
Watch out A broad label does not compensate for a panel concentrated in one economic channel.
- 2
Standardize the question
Ask the same person the same current-period question every time. Hyman's example asks how sales this week compare with what the respondent expected.
Pro tip Use the prior reading when the regular respondent is temporarily unavailable rather than introducing a different person's judgment.
Watch out Changing respondents or wording can create movement that did not occur in the business.
- 3
Create reciprocal value
Give participants the aggregate survey result and relevant economic research. This lets a company compare itself with peers while preserving the usefulness of candid participation.
Pro tip Return a consensus that helps participants judge whether they are gaining or losing share.
Watch out Do not ignore incentives that could encourage strategic or false responses.
- 4
Interpret the nominal signal
Determine whether revenue strength reflects greater activity, higher prices, or both. Use industry context such as miles driven, freight rates, passenger activity, or fares.
Pro tip Pair a revenue response with operating and pricing evidence where available.
Watch out A nominal increase can overstate real activity when prices are rising.
- 5
Triangulate official data
Compare each private reading with relevant government measures and note publication lags. Investigate agreement and divergence rather than selecting the more convenient result.
Pro tip Track whether survey changes repeatedly precede, coincide with, or contradict the official series.
Watch out Fast evidence is not automatically more accurate than slower evidence.
- 6
Update the nowcast
Combine the panel and official evidence into a present-tense view of the economy. Refresh it at the frequency supported by data and communication technology.
Pro tip State that the survey is coincident rather than predictive.
Watch out Do not turn knowledge of current conditions into false certainty about next quarter.
In the wild
Hyman's trucking survey became very strong, but the team did not treat the full increase as greater freight activity. Rate increases also contributed, so the interpretation separated miles driven from pricing before incorporating the signal into the current economic view.
→ The team retained the useful strength signal without confusing nominal revenue with physical volume.
A research team collects weekly readings from stable retail and transport respondents, then compares the changes with later retail-sales and GDP releases. When the official data arrives weeks later, the team records whether it confirmed or diverged from the private readings and adjusts how it weights each series.
→ The analyst gets an earlier current view while preserving an independent official cross-check.
Common mistakes
Rotating respondents casually
A new respondent can change the answer even when the company's condition has not changed.
Treating revenue as volume
Revenue can strengthen because prices or rates rose, so activity and price need separate interpretation.
Calling a nowcast a forecast
Hyman explicitly describes the surveys as knowledge of what is happening now, not a prediction of the next week or quarter.
Is it for you?
Best for
It is best for research teams that need a recurring, current view of business conditions across multiple industries.
Not ideal for
It is not ideal for precise long-range forecasts or situations where a stable, candid respondent panel cannot be maintained.
From the transcript
“Same question every week to the same person.”
“They're not the least bit predicted. We're not asking what do you think sales will be next week, next quarter?”
“So you're always going back and forth to try and triangulate the best picture”
From the episode
Evercore ISI Ed Hyman: Masters in Business (Audio)
Ed Hyman