How to Estimate Installed Base and Replacement Opportunity When Public Data Is Weak

Who this guide is for

This guide is for industrial equipment manufacturers, component exporters, maintenance suppliers, and B2B teams selling into an installed base. It is especially useful for pumps, compressors, production machinery, controls, HVAC, process equipment, material-handling systems, and other durable assets whose public market data are fragmented.

Contents

1 Define five linked populations 2 Choose the right estimation architecture 3 Estimate eligible sites and equipment density 4 Build age cohorts and survival assumptions 5 Convert installed base into replacement exposure 6 Find and validate public evidence 7 Quantify uncertainty and avoid false precision 8 Use the installed-base worksheet 9 Illustrative example 10 Common mistakes 11 What this does not mean 12 Common questions 13 Sources and methodology

1. Define five linked populations

Do not use "installed base" as one undifferentiated number. Define: 1 Eligible sites: facilities where the target application could exist. 2 Installed units: equipment physically present, including idle or standby units if the perimeter includes them. 3 Operating units: units currently in service. 4 Replaceable units: operating or failed units that could credibly be replaced by the offer within the planning horizon. Then define a fifth commercial population: 1 Reachable replacement opportunities: replaceable units attached to accounts the team can serve, qualify, and access. Each transition needs evidence. A food plant may be eligible, but lack the target process. A facility may have ten pumps, but only two may match the operating envelope. A 20-year-old machine may continue running because shutdown cost is higher than its maintenance burden. A replacement project may be specified by an engineering contractor rather than the plant. Write the perimeter: The model estimates [equipment class] used in [application] at [facility segments] in [geography], including [operating/standby rules] and excluding [listed configurations], as of [date].

2. Choose the right estimation architecture

Use one or more of three approaches. Bottom-up site-density model Installed base = sum of eligible sites by segment x application-fit rate by segment x units per application-fit site by segment The unit-density input must be measured only among sites that fit the application. If the available density is instead an average across all eligible sites, including zero-unit non-fit sites, do not apply the application-fit rate again. This is practical when facility counts and sample observations exist. Shipment-survival model First define cohort additions in geography as the best-supported net units added to the stated installed-base geography in each year. Reconcile domestic producer shipments, imports, exports or re-exports, used-equipment inflows and outflows, and material channel-inventory timing where applicable. A domestic shipment series is not automatically a geographic installation series. Installed base at year T = sum of each geographic cohort addition x its survival probability at T + relevant secondary units - removals not already captured by the survival assumptions For geographic cohort-addition year y: Surviving units at T = geographic cohort additions in y x survival probability at age T - y This works when long-run shipment quantities and retirement assumptions are usable. Do not subtract a documented removal if the cohort survival probability already incorporates it. Capacity-intensity model Installed base = relevant operating capacity x units per capacity unit This can help when equipment count tracks production lines, beds, building area, megawatts, throughput, or another operational denominator. Triangulate rather than averaging blindly. If the site-density model implies 40,000 units and the shipment-survival model implies 12,000, investigate the perimeter, density, shipment coverage, exports, used equipment, or survival curve.

3. Estimate eligible sites and equipment density

The U.S. Census Bureau's County Business Patterns provides annual establishment statistics by detailed industry, geography, and employment-size class. U.S. Census Bureau - County Business Patterns API Use it to build a facility denominator, not to declare application fit. Segment before assigning density: Segment dimension Why it can affect equipment density Facility size More lines, floorspace, shifts, or capacity Process type Batch, continuous, discrete, cold-chain, cleanroom, or another process Facility role Production, warehouse, laboratory, service, or office Age or retrofit status Different technology generations and layouts Region Industry mix, climate, codes, service practices Ownership Multi-site standardization and procurement model Use three density sources where possible:

• official equipment or building-characteristic surveys;

• verified facility documents, permits, asset lists, and procurement records;

• structured interviews with operators, engineers, service firms, and distributors. The Energy Information Administration's Commercial Buildings Energy Consumption Survey collects building characteristics, energy uses, and energy-using equipment. EIA - About CBECS The Manufacturing Energy Consumption Survey collects information from manufacturing establishments about energy-related characteristics, consumption, and expenditures. EIA - About MECS These surveys may provide valuable denominators or equipment context for covered categories, but should not be stretched to products they do not measure. Record the density evidence as a distribution, for example 1-2 units for small sites, 3-6 for medium sites, and 8-15 for large sites. Do not collapse it into one national average until the segment mix is applied.

4. Build age cohorts and survival assumptions

Replacement does not occur automatically at the end of an accounting life. Physical assets can remain in use, be rebuilt, moved, cannibalized, or kept as standby. The Bureau of Economic Analysis Fixed Assets tables provide estimates including net stocks, depreciation, investment, and average age for asset aggregates and detailed nonresidential assets by industry. BEA - Fixed Assets These aggregates are useful context for capital-stock patterns; they are not a product-specific equipment count or prescribed replacement life. Build a cohort table: Cohort field Required input Installation year or band Exact year or bounded age range Units installed Official, company, or modeled quantity Survival probability Low, central, high assumption by age Operating share Portion believed to remain in use Technology generation Compatibility and obsolescence context Service state Normal, degraded, rebuilt, idle, unknown Evidence quality Direct, inferred, proxy, or unknown Use a survival curve, not a cliff. If the central life is 12 years, that does not mean every unit disappears in year 12. A simple planning curve may retain most units in early years, gradually reduce the cohort through a replacement window, and preserve a long tail. The curve is a SellersGPT modeling recommendation, not an official asset-life standard.

5. Convert installed base into replacement exposure

Age is only one trigger. Create a trigger map: Trigger Evidence examples Why purchase may still not happen Failure risk downtime, repeated repair, parts scarcity repair remains cheaper Capacity constraint backlog, utilization, added shifts bottleneck is elsewhere Efficiency gap energy or labor burden payback fails budget threshold Compliance change adopted requirement, failed inspection deadline, scope, or applicability differs Product change new specification or process retrofit is sufficient Facility project expansion, relocation, renovation project is delayed or canceled Trigger Evidence examples Why purchase may still not happen Vendor risk end-of-support or long lead time buyer stockpiles parts instead Estimate annual replacement exposure, not expected sales: For an age-based model: Replacement exposure = surviving installed units x operating-or-otherwise-replaceable share x relevant age-cohort share x trigger probability range Estimate the operating-or-otherwise-replaceable share from evidence; do not assume that every surviving installed unit is active or economically replaceable. If the model starts from a directly verified trigger cohort instead of an age cohort, count the replaceable units in that cohort and do not apply the same trigger factor a second time. Then apply commercial screens separately:

• product and application fit;

• account reachability;

• specification access;

• service and installation feasibility;

• budget and timing;

• incumbent switching cost;

• unit economics. The result should be a list of account hypotheses and trigger windows, not a forecast of units won.

6. Find and validate public evidence

Use a source ladder. Level 1 - Official aggregates County Business Patterns provides establishment counts. Historical Annual Survey of Manufactures data include industry measures such as capital expenditures and product shipments; the Census Bureau states that the program transitioned to the Annual Integrated Economic Survey. U.S. Census Bureau - About the Annual Survey of Manufactures BEA, EIA, state agencies, and local permits can add asset, energy, construction, or facility context. Level 2 - Account disclosures The SEC's EDGAR search provides public access to company filings and full-text search. SEC - Search Filings Search annual and current reports for capacity, capital expenditure, facility projects, impairment, maintenance, modernization, and supply constraints. A filing describes the reporting company, not the whole market. Level 3 - Procurement and project documents SAM.gov publishes federal procurement notices including pre-solicitation, solicitation, award, and sole-source notices. SAM.gov - Contract Opportunities Specifications can reveal quantities, incumbent configurations, performance requirements, and replacement timing for public buyers. They are not representative samples unless proven otherwise. Level 4 - Facility verification Review facility pages, permits, satellite or mapping context where appropriate, product lines, engineering announcements, distributor locations, and maintenance roles. Then interview a bounded sample. For every observation, record facility, segment, date, source, equipment count or range, application, confidence, and whether the evidence can be generalized.

7. Quantify uncertainty and avoid false precision

Build low, central, and high cases from explicit variables: Variable Low Central High Eligible sites verified minimum best-supported count inclusive upper bound Fit rate conservative sample result weighted estimate plausible upper bound Units per site lower segment density median weighted density upper segment density Operating share lower verified share central survival case high survival case Trigger share hard triggers only age plus verified triggers broad exposure window Do not combine every high assumption unless that scenario is jointly plausible. Use sensitivity analysis to identify the variables that matter most. If units per large site drives most uncertainty, research large sites rather than collecting more small-site observations. Assign confidence by component:

• High: direct official or facility evidence with current scope.

• Medium: convergent evidence with a documented proxy.

• Low: sparse observations, broad classifications, or unverified assumptions. A wide range is useful when it exposes the decision risk. False precision hides it.

8. Use the installed-base worksheet

Maintain five linked tabs or tables: 1 Perimeter: product, application, geography, operating and standby rules. 2 Site universe: facility, segment, NAICS, size, fit evidence, status. 3 Density evidence: observed or reported units, source, confidence, generalization rule. 4 Cohorts: installation period, survival, operating share, technology, trigger. 5 Opportunity layer: replacement exposure, buyer path, service requirement, next evidence action. Include a model-change log. When an interview changes density from 5-8 to 3-6 units for a segment, record why, who approved the model change, and how the total range changed.

9. Illustrative example

Illustrative example. Every facility, count, rate, and conclusion below is fictional. It is not a market forecast or customer result. A compressor supplier studies refrigerated warehouses. Public data indicate 1,200 candidate facilities: 300 small, 400 medium, and 500 large. Facility verification suggests application-fit ranges of 35-50 percent for small sites, 60-75 percent for medium sites, and 80-90 percent for large sites. Observed equipment density ranges from 1-2, 3-5, and 6-10 units respectively. Calculated from fictional data: the low case is (300 x 0.35 x 1) + (400 x 0.60 x 3) + (500 x 0.80 x 6) = 3,225 units. The high case is (300 x 0.50 x 2) + (400 x 0.75 x 5) + (500 x 0.90 x 10) = 6,300 units. Independent checks suggest that 80 percent of surviving installed units in the example perimeter are operating or have failed but remain economically replaceable. The central cohort model suggests that 22 percent of that replaceable population is in an elevated age window, and only half of the age-cohort units show an observable trigger such as repeated maintenance hiring, an expansion project, or end-of-support evidence. Applied consistently to the installed-base range, that produces approximately 284-554 units of replacement exposure: 3,225 x 0.80 x 0.22 x 0.50 = 283.8 and 6,300 x 0.80 x 0.22 x 0.50 = 554.4, rounded to whole units. It does not produce expected orders. A sample of 60 facilities identifies only 18 accounts with confirmed application fit and a plausible service path. The GTM plan starts with those accounts and uses every conversation to refine density, operating status, replaceability, incumbent, age, and trigger assumptions.

10. Common mistakes

Multiplying every establishment by one average Segment by facility size, process, and application. One mean can overstate small sites and understate large ones. Dividing installed base by useful life That assumes uniform age, deterministic retirement, and steady replacement. Build cohorts and survival instead. Confusing exposure with sales An aging unit may be repaired, removed, or replaced by an incumbent. Apply access and switching tests. Treating accounting age as physical age BEA and company accounting measures are contextual. They do not reveal the installation date of a named machine. Generalizing from visible large facilities Large sites publish more information and can bias density upward. Weight observations to the actual site universe. Hiding model revisions An estimate should improve. Preserve its versions so later users can see why it changed.

11. What this does not mean

• The framework does not create an official installed-base statistic.

• An old unit is not necessarily obsolete, unsafe, or ready to replace.

• A replacement trigger does not prove budget, authority, or purchase timing.

• Public procurement and public-company evidence may not represent private SMEs.

• The model does not predict market share, leads, revenue, or order probability.

• Equipment life, compliance, and safety decisions require qualified technical and regulatory review.

12. Common questions

How many facility observations are enough? There is no universal number. Sample every material segment and continue until additional observations no longer change the range enough to alter the decision. Report coverage and bias. Can distributor estimates be used? Yes, as attributed primary evidence. Ask for the geography, application, customer type, time frame, and basis. Compare several independent views and do not relabel them official facts. What if shipment history is unavailable? Use the site-density model, cohort evidence from facility ages or projects, and a broad survival range. Keep the missing shipment history in the limitation register. Should standby units be included? Only if the commercial offer can replace or service them and the perimeter explicitly includes them. Keep operating and standby populations separate.

13. Sources and methodology

Sources 1 U.S. Bureau of Economic Analysis - Fixed Assets. Used for the availability of fixed-asset stock, depreciation, investment, and average-age estimates. Accessed August 24, 2026. 2 U.S. Energy Information Administration - About the Commercial Buildings Energy Consumption Survey. Used for the scope of building-characteristic and energy-using-equipment data. Accessed August 24, 2026. 3 U.S. Energy Information Administration - About the Manufacturing Energy Consumption Survey. Used for the survey scope covering manufacturing establishments and energy-related characteristics. Accessed August 24, 2026. 4 U.S. Census Bureau - County Business Patterns API. Used for establishment counts by detailed industry, geography, and size class. Accessed August 24, 2026. 5 U.S. Census Bureau - About the Annual Survey of Manufactures. Used for the historical availability of manufacturing capital-expenditure, output, inventory, and product-shipment measures, and the transition to the Annual Integrated Economic Survey. Accessed August 24, 2026. 6 U.S. Securities and Exchange Commission - Search Filings. Used for access to company filings and full-text disclosure research. Accessed August 24, 2026. 7 SAM.gov - Contract Opportunities. Used for the availability and types of U.S. federal procurement notices. Accessed August 24, 2026. SellersGPT analysis and framework The population ladder, three estimation architectures, cohort design, replacement-exposure equation, source ladder, confidence labels, and worksheet are original SellersGPT editorial tools. The example is fictional and no actual installed-base or market forecast is stated. Update policy Review the model whenever official establishment data, asset surveys, facility evidence, shipment history, or replacement observations change. Revalidate survival and density assumptions at least annually for an active market.