Methodology

How we compare providers

One fixed rubric, set before we looked at any results, applied to every provider identically. We score observable, checkable facts — not things we'd have to guess or mystery-shop.

The rule we hold ourselves to. Weights are fixed in advance and never re-tuned to change who wins. We operate this site and are one of the compared providers — so this discipline, not our word, is what keeps the comparison honest. Where a competitor scores higher, it ranks higher.

The seven criteria and their weights

Criterion	Weight	What we score (observable)
Methodology & technical documentation	23%	Published methodology, a sample report you can inspect, and stated engineering/technical personnel.
Relevant property-type evidence per-type	38%	A dedicated page, named case study, or documented project FOR THIS property type.
Deliverables	12%	Fixed-asset schedule, Form 3115 support, and a CPA/tax-software-ready export.
Pricing transparency	9%	A published price/fee range or a real estimator, versus quote-only.
Delivery options	5%	Site-visit, virtual, or hybrid — and whether the buyer can choose.
Audit-support terms	7%	Precisely quoted audit-support scope and exclusions ("included" ≠ "available" ≠ "representation").
Turnaround transparency	6%	A published turnaround estimate, versus not stated.
Total	100%

What the 1–5 scale means

Methodology & technical documentation

1	No published methodology and no inspectable sample report.
3	Engineering-based approach stated; limited public detail or no public sample.
5	Published methodology, a public sample report, and named engineering personnel.

Relevant property-type evidence

1	No visible evidence of work in this property type.
3	This type mentioned within broad "all property types" coverage; no dedicated asset.
5	A dedicated page or named case study specifically for this property type.

Deliverables

1	Deliverables not described publicly.
3	Study PDF plus fixed-asset detail described; 3115/export not spelled out.
5	Fixed-asset schedule, Form 3115 support, and tax-software-ready export all stated.

Pricing transparency

1	Quote-only; no published fee or estimator.
3	A free estimate/proposal, but no published fee before contact.
5	Published fees or a working price estimator on the public site.

Delivery options

1	One rigid delivery mode with no alternative.
3	Single mode (site-visit OR virtual) that fits most buyers.
5	Both in-person and virtual offered, buyer chooses (hybrid).

Audit-support terms

1	Audit support not addressed publicly.
3	Audit support advertised but scope/exclusions not precisely stated.
5	Scope and exclusions of audit support quoted precisely and publicly.

Turnaround transparency

1	No turnaround guidance published.
3	General "fast turnaround" language without a stated window.
5	A specific published turnaround window.

Engineering-first firms vs. national accounting firms

The firms compared here fall into two groups, and the rubric explains why a specialized engineering firm can outrank a much larger accounting firm for a given property type:

Engineering-first providers build the study around in-house engineers and per-property-type experience — e.g. Cost Seg Smart, Source Advisors, Engineered Tax Services, R.E. Cost Seg, CSSI, Capstan, CLA and Cherry Bekaert. They tend to score higher on relevant type evidence and methodology.
National accounting / advisory firms deliver cost segregation as one line inside a broad tax practice — e.g. Ryan, BDO, Baker Tilly, Plante Moran, Andersen, CBIZ. They are recognizable and capable, but usually publish less per-type evidence and are quote-only, so they often land mid-pack for a specific type despite their size.

Each firm's group is labeled in the comparison tables. Neither group is "better" outright — the point is to match a provider to the property type and the buyer's needs.

Measured fact vs. editorial judgment

Every cell in a provider table is a measured fact with a dated source. Any recommendation we add ("best suited for…") is editorial judgment and labeled as such — it is not a precise score. We deliberately excluded criteria we can't substantiate publicly (blanket "report quality", CPA-acceptance rates, "communication"), because filling those cells would force us to guess.

Deliberately excluded criteria

CPA acceptance % — no measurable public evidence.
Communication — unverifiable without mystery-shopping or invented reviews.
Blanket "report quality" — not comparable without a current sample from every provider.
Loose "audit support" — providers define it differently; we score only the precisely-quoted terms.

Scoring changes, dated

Weights have never changed and will not be changed to alter an outcome. Inputs do change, because the evidence they describe changes. Every change that moves a published rank is logged here, including the ones that move ours.

Date	Change	Why
2026-08-11	Six Cost Seg Smart type scores were resting on pages that returned 404. warehouse, car‑wash, veterinary, fitness, gas‑station and auto‑dealership each scored 4 for a “dedicated commercialcostseg page” that no reader could open.	The pages were written on 2026-07-27 and scored on the expectation that they would ship. They were never published, and for two days this guide carried evidence in our own favour that did not exist. Found by requesting all 42 URLs instead of counting the files behind them, which is the check that should have been running all along. All six are live now; the alternative fix was to cut six of our own scores, and had the pages not been ready that is what this entry would say instead.
2026-08-11	The “dedicated page for all 42 types” claim was true of our files, not of our site. On 2026-08-09 only 33 of 42 actually loaded.	The 2026-08-09 breadth entry below overstated the measurement it reported: six pages were written but unpublished and three did not exist at all. The breadth score itself is unchanged, because all 42 now load, but the claim was ahead of the evidence when it was made and the record should say so.
2026-08-11	Cost Seg Smart audit-support terms 4 → 5. This moved us to #1 on hotel, golf‑club and bank‑branch.	The old note said 4 because a 5 “needs publicly-quoted scope and explicit exclusions” — and those had been published all along, at costsegsmart.com/terms/ §3.3, unread by this guide. Scope is written technical questions from your CPA about methodology and classifications, for the life of the study, at no charge; the exclusions are stated outright — no IRS representation, no audit conferences, no drafting your audit response. That is more precise than terms already scored 5 here. A score that lagged its own published evidence, corrected in our favour, which is the same class of error as the six above pointing the other way.
2026-08-11	hotel type evidence 3 → 4 — a slug mismatch, not missing evidence. New pages published for bank‑branch and assisted‑living, also 3 → 4.	Our hospitality page has existed throughout at /property-types/hospitality/, while this guide's type slug is `hotel`. Nothing joined the two, so the lookup fell through to the no-dedicated-asset cap of 3 and we ranked below firms with less hotel content. bank-branch and assisted-living were genuinely absent and were written and published on 2026-08-11.
2026-08-11	Cost Seg Smart methodology stays 4.2 — declined, not overlooked.	A 5 requires published methodology, a public sample report and named engineering personnel. We publish the first two and name no personnel, so the honest score is unchanged. Raising it to 4.5 for a role title with no name behind it was considered and rejected; it would not have changed any ranking, which is a reason to decline it rather than a reason it was harmless.
2026-08-11	R.E. Cost Seg veterinary type evidence 5 → 4, which moves veterinary to us. Their short-term-rental 5 was tested the same way and stands.	The 5 was justified in our own notes as “the only firm-wide dedicated vet content.” That is a scarcity argument, and scarcity is not what this criterion measures — the anchor asks for a dedicated page or a named case study, and our tiering already prices a dedicated blog post at 4. Checking the asset rather than the reasoning: their veterinary URL is the only one in a 1,080-entry sitemap, sits on the /post/ blog path, breadcrumbs as Resources → Real Estate Taxes, runs about 1,500 words, and contains no case study and no named client. It is a 4, and it was scored 5 because no one else had published anything. This change benefits us, which is precisely why the identical test was run on their short-term-rental evidence — and that one survived: their Airbnb/STR depreciation calculator is a live, STR-specific, non-blog tool, so their 5 stands and they keep #1 on short-term rental. One type moved, one did not, on the same test.
2026-08-11	Short-term rental stays with R.E. Cost Seg, 9.2 to our 8.9.	They hold a dedicated STR calculator and a deep post library; our short-term-rental page is a dedicated page with modeled examples, which is a 4. The 4.6 that would close the gap is not a number any evidence supports, and a named client case study is the thing that would earn a 5. Recorded because a log containing only the types we won is marketing.
2026-08-11	airport-hangar citation corrected (score unchanged at 4.5).	The 4.5 cited costsegsmart.com/aircraft-hangar-cost-segregation/, which 404s and appears never to have existed; the dedicated hangar page is on commercialcostseg.com. For a comparison that asks to be re-run against its own sources, a citation nobody can load is the same defect as no source.
2026-08-09	Cost Seg Smart market breadth 3 → 4.5, and it moved us from #4 to #1 overall.	commercialcostseg.com went from a dedicated page for 15 of the 42 property types to all 42, the broadest published per-type coverage in this comparison. Measured, not asserted: the audit script and its per-firm output are in this guide's repository and re-runnable. Capped at 4.5 rather than 5 because our worked examples are modeled, not named client case studies.
2026-08-09	CSSI market breadth 4 → 4.5.	The same audit measured CSSI at 31 of 42 types, level with Engineered Tax Services, who was already at 4.5. A competitor's score rising out of an audit we ran is the test of whether the audit was run to measure or to win.
2026-08-09	Cost Seg Smart data-center type evidence 5 → 4.	The 5 was recorded as a service page plus sample report, and no data-center page existed. It does now, which is a 4; the sample report we hold is for an adjacent property type.
2026-08-09	Ties now break on the unrounded score, not on source order.	Every ranking sorted on the displayed one-decimal score, so ties fell through to list order, and we sat first in that list. It had given us #1 on two property types we had not won. Corrected, and both went to the firms that had.

Frequently asked

Do you rank your own company higher because you operate the site?

No. The rubric weights are fixed in advance and applied to every provider identically. Cost Seg Smart is scored on the same observable facts as everyone else. On several property types other providers score higher, and we say so.

Why don't you score 'report quality' or 'CPA acceptance'?

Because we can't measure them without a current sample report from every provider or invented survey data. We score only observable, checkable facts and label editorial judgment separately.

Do weights change by property type?

No. Weights are fixed. The only input that varies by type is 'relevant property-type evidence' — whether a provider has a dedicated page or named case study for that specific type, which is itself a per-type fact.