Use purchasing managers’ indexes (PMIs) and other leading indicators to inform a business forecast—not to replace it. These measures provide outside context about economic direction and timing; your company’s orders, customer behavior, pricing, capacity and costs should remain the direct drivers of its forecast.
What PMI can—and cannot—tell you
A PMI is a monthly, survey-based diffusion measure. Business executives report whether specified conditions rose, fell or stayed unchanged compared with the previous month; the index summarizes the balance of those responses. For S&P Global’s PMI, 50 indicates no net change, a reading above 50 generally indicates expansion, and a reading below 50 generally indicates contraction relative to the prior month. It is not a percentage change in output or a forecast of your company’s sales growth. S&P Global explains its PMI methodology and products.
PMI’s value is its timely view of direction and breadth. Because it is released ahead of many comparable official statistics, it can help monitor the economy or inform a GDP nowcast. An economy-wide nowcast is not a firm-specific forecast. S&P Global’s PMI FAQ and its guide to interpreting the PMI output index describe these uses.
Choose a series that fits your exposure
Use a manufacturing PMI for relevant manufacturing exposure, a services business activity index for services, and an appropriate composite when your business spans both. Match the series’ geography and industry to the markets that generate revenue or determine costs. A national headline may be too broad to describe a niche segment or a company operating in another country.
#1 Best Overall
Read the components, not just the headline
Sub-indices can help explain what is moving: new orders and new business may indicate demand direction; output or activity describes current volume; employment and backlogs can add context about capacity; and prices, supplier delivery times and inventories may inform cost or supply scenarios. These are possible connections to investigate, not automatic causal links to company results.
For S&P Global’s manufacturing PMI calculation, the FAQ lists weights of 30% for new orders, 25% for output, 20% for employment, 15% for supplier delivery times (inverted), and 10% for stocks of purchases. Those are index-calculation weights, not suitable default weights for a company forecast. S&P Global’s services headline is a Services Business Activity Index based on a business-activity question. See the PMI FAQ for these methodology details.
Build the forecast from company drivers
Start with the business outcome you need to forecast, then use external indicators only where there is a plausible mechanism linking them to a company assumption. For example, external demand conditions could inform a pipeline or order scenario; input-price signals could prompt a cost sensitivity. Keep actual company data—such as its order book, customer retention, conversion rates, pricing, staffing, capacity and costs—as the direct forecast drivers.
- Define the forecast question. Specify the outcome, time horizon, geography, business segment and update cadence. Decide whether an indicator is intended to inform demand, price or cost, staffing, investment, or downside risk.
- Map the company’s exposure. Identify the sectors and regions that produce revenue or shape costs. Select PMI series accordingly; do not assume an aggregate index precisely represents a specialized business.
- Inspect relevant sub-indices. Record which components appear relevant to the question—orders, activity, backlogs, employment, prices, delivery times or inventories—and treat the connection to company performance as a hypothesis to test.
- Add a complementary cross-check. For US business-cycle context, The Conference Board’s Leading Economic Index (LEI) is designed to signal turning points, while its Coincident Economic Index (CEI) tracks current conditions. The LEI combines multiple components, so check whether it overlaps with other indicators before treating the measures as independent confirmations. The Conference Board’s US indicators page provides current release and methodology context.
- Record timing and data vintage. Note the survey’s reference month, the publication date, and whether a reading is flash or final. Preserve the version of each indicator used for each forecast so you can later compare like with like. Early publication helps with timeliness; it does not remove uncertainty.
- Translate a signal into a labeled assumption. Document the business mechanism, the company driver affected, and the size or range of the planned adjustment. Keep the published index observation separate from the analyst’s judgment about its company-specific effect.
- Build scenarios and test the relationship. Create a base case and plausible upside and downside sensitivities. Compare historical indicator movements with the company outcome at the relevant horizon. Check whether any relationship is stable, differs by segment, or breaks in unusual periods; correlation alone does not establish causation.
- Review on a fixed cadence. When new releases arrive, record which assumptions changed and why. Compare forecast errors with actual results, and retain an audit trail of the series, transformations, assumptions, owners and decision dates.
How to compare leading indicators
More indicators do not automatically make a better forecast. Choose measures connected to a decision or business driver, and compare them on the dimensions that affect whether they add useful information.
Free tools Windows power users keep installed
One-click scans. No signup required.
- What they measure: survey direction, actual activity, orders, employment, prices, financial conditions or a composite.
- Timing and lead or lag: when the series is released and any stated lead horizon. Do not assume the same timing applies across industries, geographies or companies.
- Coverage: whether the geography and sector match your company’s revenue or cost exposure.
- Frequency and publication lag: an early monthly survey may arrive before official output data; a slower measure may offer a later, more complete reading.
- Preliminary status and revisions: distinguish flash estimates from later updates and preserve the data vintage used in each forecast.
- Overlap and independence: identify shared or overlapping components so correlated measures are not counted as multiple independent confirmations.
- Actionability: favor indicators that could change a forecast assumption or decision over a large dashboard with no defined use.
Limits to keep in the forecast
- A diffusion reading is not a growth rate. A PMI level does not state the percentage change in your output, sales or profit.
- The economy is not the company. Market share, product mix, execution, customer concentration and contracts can make a business outperform or lag its sector.
- There is no established universal PMI-to-sales conversion. The cited indicator sources do not provide a general equation for converting PMI into company revenue, profit or cash flow. Estimate any relationship using company history and label it as an assumption.
- A headline can hide mixed conditions. Components such as orders, employment, delivery times, inventories and prices may move in different directions.
- Lead times are estimates, not guarantees. The Conference Board describes an approximate seven-month lead time for the US LEI’s anticipation of business-cycle turning points. That estimate is specific to this index and geography; it is not a guaranteed horizon for a company forecast. See The Conference Board’s US LEI information.
- Geography and sector matter. A US manufacturing reading does not automatically describe a services company in another country.
Keep the evidence and the assumption distinct
A sound forecast shows which facts came from released indicators and which decisions came from company analysis. State the series, geography, reference month, release date and preliminary status; then explain which company driver it informs and how you tested that link. That makes it easier to update the forecast when a new release arrives—or when actual company results show the assumed relationship was not useful.
Quick Recap
Best Value
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
