Mapping an installed base without guessing equipment brands

What does an installed-base map actually measure?

Map the installed base from a unit register with evidence for each location, identity and current status. A map pin alone does not tell a distributor which product family to support or a technician which parts to carry.

Keep three deliverables separate: the visible equipment inventory, the verified manufacturer inventory, and the qualified account list. The first supports where to investigate. The second supports product coverage and replacement research. The third adds an identified decision maker, a confirmed need, and permission to discuss the work. Moving a row between them requires new evidence.

Use the same register for generator yards, rooftop plant and interior equipment confirmed through records or a visit. Set the business purpose first: dealer coverage, parts planning, owner inventory or a survey route. Then specify the evidence needed for that decision.

Define the territory and counting unit

Write the boundary before counting anything. Name the counties, service radius, or supplied polygon; the equipment classes; the observation cutoff; and the inclusion rules. Decide whether proposed installations, removed units, warehouse stock, rental equipment, and inaccessible roofs belong in separate queues. Keep an exclusion reason so another analyst can reproduce the boundary.

Choose the unit of count explicitly. A campus, building, equipment system, packaged assembly, and component are different objects. For this workflow, give each assembly its own asset ID and link components underneath it. Define a cooling-tower system and its cells separately. Define a generator set separately from its engine, alternator, enclosure, and fuel tank. Do not let a component label become the assembly manufacturer.

Keep site counts and equipment counts in different columns. One campus can have several buildings and many equipment positions. One property owner can have several campuses. State which entity receives the territory assignment so two dealers do not receive the same account merely because its equipment crosses a parcel line.

PorchRocket's Asset Search connects teams working on installed-base mapping method to sites with visible outdoor installed equipment and the owner of record for each find.

What can an image establish?

An image can support a dated observation of visible form and location. Record the equipment class you can defend, the counted objects, and the features that support the classification. If the view supports only rooftop mechanical equipment, use that label. Do not fill in rooftop unit, chiller, capacity, age, or brand to make the spreadsheet look finished.

The Centers for Disease Control and Prevention (CDC) identifies cooling-tower search limits that matter here: side-mounted fans can be hard to see, devices can be hidden, and maps can be outdated. CDC combines imagery with ground searching, industry knowledge, and permitting or registration records. Its guidance supports finding possible devices. It does not provide a method for assigning equipment brands from cabinet appearance.

Save the image acquisition date separately from the day you reviewed it. Mark screened equipment, heavy shadow, roof obstructions, and ambiguous objects as unresolved. A clear view of a housing does not establish whether the unit operates, what it serves, or whether the owner wants service. A roof with no visible equipment needs a visibility assessment before it becomes a zero in your market count.

The evidence ladder

Use the ladder below as a release rule for individual fields. It is a recommended working method, not a certification standard. Each step answers a different question. A permit may document a proposed model more clearly than an owner's recollection, while a current field photo may show that the proposed unit was substituted.

Keep the underlying claims when sources disagree. Record proposed model, observed model, and owner-reported model separately until someone resolves the conflict. An overall confidence score should never conceal an unsupported manufacturer field. Use unknown, reported, verified, and conflicting as explicit states, with the supporting evidence beside each one.

Before scaling the pilot, have a second reviewer classify a selection of the same evidence without seeing the first decisions. Include ambiguous views and rejected objects. Compare the disputed fields and revise the written rules before processing more sites. Agreement between reviewers measures consistency of your method; it does not independently establish that a hidden model or brand is correct.

How do you attach the right site and owner?

Create your own site ID, then retain the identifiers from every source. Keep the parcel, building identifier, street address, campus name, and equipment location. Address spelling can change while the equipment stays put. Coordinates help place a candidate, but the nearest address is a hypothesis until the site boundary and serving building are checked.

The Environmental Protection Agency (EPA) Facility Registry Service uses a Registry ID across multiple EPA systems. Where an EPA facility appears in your research, retain that identifier as a source key. Treat it as a facility match, not as a serial number or an equipment count. Preserve other jurisdiction identifiers alongside it rather than replacing them.

Separate owner of record, operator, tenant, maintenance provider, and purchasing contact in your worksheet. Verify each role through an appropriate record or direct confirmation. A property match does not establish who controls a replacement purchase. For a shared plant, ask which buildings it serves and who approves access before attaching its full equipment count to one tenant account.

How do you verify the manufacturer and model?

Tie a readable assembly nameplate to a specific unit, then reconcile it with the owner register. Take an establishing photo showing the asset tag and location, followed by the plate detail. Transcribe the model and serial exactly. Keep punctuation and leading zeros. If a character is unreadable, preserve the uncertainty and request another image instead of completing the model from memory.

Keep the observed brand, the original equipment manufacturer where documented, and any normalized corporate parent in separate fields. A current corporate relationship does not rewrite the label on an older machine. Store the reference and date behind any normalization. A replacement motor or controller likewise needs its own component record rather than overwriting the parent assembly.

The Air-Conditioning, Heating, and Refrigeration Institute (AHRI) describes its directory as a source of certified product performance. Use it to check product identity and ratings after obtaining the model. AHRI's support guidance states that its database does not capture equipment serial numbers. Therefore a directory result cannot be your evidence that a particular serialized machine is installed at a particular site. Keep the location proof with the field record.

What do permits and public inventories prove?

Read a permit record as evidence about a defined project and its recorded status. Capture the issuing authority, application or permit number, scope, status, document revision, and relevant dates. Where you can obtain a schedule or completion record, match its equipment designation to the observed unit. An approved selection belongs in a proposed-equipment field until the installation is reconciled.

Ask the owner for the current equipment register, commissioning records, replacement history, and any approved substitutions. Record whether each document is proposed, as built, or maintained as an operating inventory. A purchase record supports acquisition. Your installed-base register still needs to connect that purchase to equipment at the location.

Check a dataset's coverage before treating a missing row as missing equipment. EIA-860 covers existing and proposed plants with at least 1 MW of combined nameplate capacity where the generators or their host facility connect to the local or regional grid and can draw or deliver power. Its detailed data separate operable, proposed, and retired or canceled records; the last group is not a complete retirement history. Keep the reporting year and status with each row. This is not a census of every commercial standby generator.

A small verified public-data sample

The NYC Cooling Tower System Inspection Results CSV downloaded November 4, 2026 supplies this counting check. Filter System_ID to the three values below, group by BIN and System_ID, count rows and distinct Inspection_date values, and retain the exported Status. BIN is the building identifier; System_ID is the registered system identifier.

The extract contains 110 rows for these three systems. The export has no manufacturer or model columns. Keep those fields unknown. Inspection rows are not equipment units, and the exported status is not a new field inspection. Use this selected sample to practice data reconciliation.

Use the linked CSV and column definitions to repeat the check. The live export can change, so retain your downloaded version and retrieval time when using this method commercially. Do not carry an inactive system into the active installed-base total merely because its inspection history remains searchable.

Field verification and reconciliation

Arrange the walkdown with the responsible site contact. Send the candidate sheet first, including the equipment positions and the questions you need answered. Ask for a current register and an escort who can distinguish installed, spare, disconnected, and removed equipment. Agree on photography and access before the visit so the crew knows what evidence it can collect.

This is an identity survey. Keep functional testing, panel opening, equipment operation, and maintenance in separately authorized scopes. Follow the site's access and safety procedures. If the plate cannot be reached within the approved survey, record the limitation. An unreadable nameplate is a research exception to close, not a reason to improvise access around operating plant.

Before releasing the file, reconcile every candidate position as verified equipment, rejected object, duplicate, removed equipment, or unresolved. Keep the original sighting ID when a unit is replaced and create a new asset identity linked to the predecessor. Record who resolved each discrepancy and from which evidence. The next reviewer should be able to follow the decision without calling the surveyor.

Close negative findings with the same care. If a candidate is absent, record the area actually checked and the evidence supporting removal or rejection. If the escort could not reach the roof, the result is access unresolved. That distinction prevents a failed visit from silently shrinking the installed-base count.

What to document

Attach a source, date and reviewer to each material field. Keep acquisition, publication, retrieval and verification dates separate. Assign missing evidence to someone; a fresh export can still contain old observations.

Use the worksheet below as the minimum handoff. Store evidence files under stable references and keep a change history. A customer relationship management (CRM) system can receive the site and account summary; the detailed evidence register should retain the equipment identities and conflicts behind it. Share only the fields the receiving team needs for its task.

How much should a pilot verification cost?

Price a pilot by the evidence required, travel, and access. Anvilfield research models a typical internal inventory pilot. 24 screened sites contain 60 distinct candidate equipment positions. 12 sites allow visits, covering 30 positions. Visits return 24 complete unit identities; six visited positions remain unresolved and 30 positions are unvisited.

One analyst needs 12 desktop hours, four coordination hours, and four reconciliation hours at a loaded internal rate of $65 per hour. One technician spends 1.5 survey hours plus 0.5 travel hour at each of 12 sites, totaling 24 hours at $95 per hour. That technician rate includes ordinary vehicle and survey-tool costs. Available owner escorts add no charge, every visit occurs as scheduled, and no repeat trip is needed.

The base budget is $3,580, or $149.17 per completed unit identity after rounding cents. A discretionary 20 percent reserve adds $716, giving a $4,296 authorization limit. Paid data, software, access fees, permits, lifts, taxes, testing, repairs, and sales margin are excluded. Add them explicitly if required; price each separately from the reserve.

The scope also matters: only 24 of 60 candidate positions have complete identities. Do not present the pilot as 60 verified units. If access falls through or a second technician is required, recalculate hours and cost before expanding the survey. Use the pilot to learn the real completion rate for each site class.

How do you calculate coverage without inventing market share?

Report completion against the correct denominator. In the typical pilot, 24 complete identities divided by 60 candidate positions gives 40 percent completion of the candidate queue. It does not establish 40 percent coverage of all installed equipment in the territory. Your screening could have missed hidden units, and some remaining candidates could be rejected later.

Using the same 24 verified identities in a typical product-line breakdown, 14 belong to your product line and ten to other lines. The 58.3 percent figure from 14 divided by 24 describes the verified subset only. Because access was selective, that subset cannot support a territory market-share claim. Report 14 confirmed units, ten other confirmed identities, and 36 unresolved candidate positions.

To estimate a wider population, define the site sampling frame, selection method, missing-site treatment, and uncertainty before surveying. Include sites with poor visibility and denied access in the reporting, rather than discarding them. If those conditions cannot be met, deliver a documented inventory and a gap list. That is still useful for territory planning without an unsupported percentage.

What should change monitoring trigger?

A changed exterior view should trigger a review task. Compare the acquisition dates and whether the two views show the same area clearly. Record possible addition, removal, relocation, or obstruction. Then ask the owner or project team what changed. A roof replacement, temporary equipment, or a different viewing angle can otherwise be mistaken for an installed-base change.

Set the refresh interval around the decision and available evidence. For example, your team could review active construction quarterly and established service accounts annually as an internal policy. Set review intervals to fit regulatory requirements and image availability. A replacement notification or contradictory field record should reopen the relevant identity immediately.

Keep superseded rows and add a dated reason when present status changes. Link a replacement to its predecessor so the dealer can trace old serial numbers in the service history.

Common mistakes

The weak note says: familiar cabinet, probably our brand, likely due for replacement. The useful note says: exterior mechanical assembly at candidate position C-17; class unresolved; image date recorded; nameplate not obtained; owner inventory requested. The useful note gives the next person a job they can finish.

Other errors follow the same pattern: copying a proposed model into the installed field, counting fans as whole machines, treating an old inspection as current operation, and assigning an entire shared plant to the nearest building. Make the counting and status rules visible beside the totals.

Ask the account team to verify service need, budget and purchasing route. Equipment identity alone establishes none of those. Hand over the confirmed units and the unanswered commercial questions.

Field checklist

Sources and verification limits

The primary sources below were checked for this November 4, 2026 edition. CDC's identification page shows an October 19, 2026 update. AHRI's support article shows a November 30, 2023 modification. Other references are live pages or exports; retain the version used for your own inventory. The column definitions establish the keys used in the public-data example.

These sources support identification limits, product-reference limits, facility matching, and the sample calculation. No site visit, private owner register, or equipment nameplate was obtained for this article. The public-data sample verifies exported records only. No adoption rule, equipment condition, or buying intent is inferred from the map.

Units, terms, and definitions

Installed base

Equipment documented as installed within a stated territory, equipment scope, and observation period.

Candidate position

A mapped location awaiting reconciliation as equipment, a rejected object, a duplicate, a removal, or an unresolved case.

Verified identity

An assembly identity tied to its physical location and supported by the evidence required for the project.

Counting unit

The explicitly chosen object being totaled, such as a site, system, assembly, or component.

Unknown

A field whose required evidence is missing; retain it as a visible state rather than assigning a likely value.

Methodology

Counts, prices, work rates, schedules and outcomes in typical cases are inputs to the Anvilfield research model. Calculations use the stated scope, formulas, cost inclusions and exclusions. Cited public records and technical documents support their attributed facts. Public-data examples are selected for data handling; they establish neither representative market coverage nor a service need.

Related tools

FAQ

People also ask

• What is a data center PM program?
• What is data center commissioning?
• How do you start up an RTU?
• How does a cooling tower work?
• What are range and approach on a cooling tower?