How we test DAM and photo management software
Every score on PhotoLib comes from the same repeatable process. This page documents it — so you can judge our judgments.
The test library
All tools ingest the same 25,000-image corpus: RAW from six camera brands, layered PSDs, JPEG/TIFF exports, video clips, PDFs and CAD samples, spanning 15 years of real shoots with deliberately messy, partially missing metadata — because that's what real archives look like. Top scorers additionally face a 200,000-file stress catalog. Team features are tested with real pilot groups, not solo clicking.
The 38-point scorecard
Six weighted areas: search (speed, filters, saved searches — 20%), metadata (IPTC/XMP/EXIF round-trip fidelity, controlled vocabulary, batch operations — 25%), speed (indexing, browsing, concurrent load — 15%), permissions & collaboration (roles, versioning, sharing — 15%), integrations & formats (10%), and value (capability per dollar at three team sizes — 15%). Sub-scores you see on review pages (ease of use, features, value, support) aggregate these 38 checks.
The round-trip test deserves its own sentence: we export 100 tagged assets, re-import them into a clean system, and count surviving fields against the IPTC Photo Metadata standard. Nothing predicts long-term satisfaction better.
Timing rules
Identical hardware for every timed run: Windows Server 2022 VM (8 vCPU, 32 GB), gigabit LAN to a Synology DS1522+, standardized query set, three runs, median reported. Cloud tools get the same query set from the same connection. Numbers published on this site (e.g. "0.3s on 25k assets") are these medians — from our June 2026 cycle unless labeled otherwise.
Pricing and tiers
Most DAM vendors quote individually, so we don't print exact prices that would be wrong by Thursday. Tables use tiers — $ budget, $$ mid-range, $$$ premium — calibrated from public rate cards where they exist and verified customer reports on G2 and Capterra where they don't. Every commercial page states its verification date.
Independence rules
Three rules, no exceptions: vendors can't pay for placement or scores; affiliate status is checked only after the scorecard is complete; and we buy or trial software through normal channels — when a vendor provides an NFR license, the review says so. Scores expire: full re-tests run yearly and after major releases, and every page shows its "Updated" date.
When we haven’t tested something
Some questions matter to buyers and cannot be answered from a test library. We have not run a penetration test against a vendor's infrastructure, audited a SOC 2 report, or verified a compliance certificate with the auditor who issued it. Pretending otherwise would be the easiest lie on this site to tell and the hardest for a reader to catch.
So we label the origin of every claim, and we use exactly three labels:
| Label | What it means | Can it carry a PhotoLib score? |
|---|---|---|
| PhotoLib tested | We measured it ourselves, on our own library, on a stated date. Search timings, metadata round-trip percentages, indexing speeds, setup times. | Yes. Only these do. |
| Vendor documentation | Taken from a vendor's own specification, trust centre, pricing page or support docs. Linked and dated. Not independently verified by us. | No. |
| Verified customer reports | Drawn from verified-purchaser reviews on G2 and Capterra, or from reference customers we interviewed. Useful for price bands and rollout timelines. Not independently verified by us. | No. |
Rules that follow from this, and that we hold ourselves to: a page built mainly on the second and third labels says so in a box at the top, before the reader has invested any attention. Vendor and customer-report claims never receive a PhotoLib score, a star rating, or the words “we measured.” And if a vendor claim cannot be traced to a public document we can link and date, we drop the claim rather than repeat it.
This lets us cover subjects our test library genuinely cannot reach — security certifications, compliance regimes, enterprise contract terms — without quietly borrowing the authority of testing we never did. A page that tells you it did not test something is worth more than one that leaves you guessing.
Limitations, honestly
We can't replicate every environment — a 50-editor newsroom will hit loads our pilot groups don't. Enterprise tools tested in sandbox tenants (noted per review) reflect vendor-configured environments plus customer reference interviews. Where confidence is limited, the review says so in the text rather than hiding behind the number.