Buyer's guide: Choosing and implementing artwork archive software for small galleries

Most galleries don't buy archive software because they did careful market research. They buy it after something goes wrong. A collector asks for the provenance chain on a piece from 2019, and it takes three days to track down because the records are split across a spreadsheet, two email threads, and someone's memory. Or an insurance renewal requires a valuation report, and nobody can pull clean images with dimensions and condition notes from the same place.

By the time you're actively shopping, you already know you need something better than folders and spreadsheets. What you usually don't have is a framework for choosing between the two dozen products that all look identical on their homepages and all promise to "centralize your collection."

This guide is about that decision — how to evaluate, procure, and actually roll out artwork archive software without buying the wrong thing or spending six months migrating data into a system nobody ends up using.

## Start with the failure you're trying to prevent, not the feature list

Vendors sell features. You should buy against risks. Before you look at a single demo, write down the two or three scenarios that made you start looking. They almost always fall into a small set of categories:

1. 1
   Retrieval failure — you can't find records, images, or documents fast enough when a collector, insurer, or lender asks.
2. 2
   Provenance and rights gaps — you can't reconstruct ownership history, exhibition history, or licensing terms with any confidence.
3. 3
   Loss of institutional memory — a founder or long-time registrar leaves, and half the knowledge walks out with them.
4. 4
   Duplicate and conflicting records — the same work exists in three places with three different prices and two different titles.

This matters because the "best" archive tool for a gallery whose main pain is rights and licensing looks completely different from the best tool for a gallery drowning in inventory duplication. If you don't name the failure first, you'll get seduced by whatever demo is flashiest.

Galleries that buy on features tend to switch systems within two years. Galleries that buy against a named risk tend to stay, because they can actually measure whether the tool solved the thing.

## The prioritized feature checklist (in the order that actually matters)

Not all features carry equal weight. Here's the priority order that holds up for small galleries, roughly from "non-negotiable" down to "nice if it's there."

Tier 1 — Must have or walk away:

1. 1
   [ ] Structured records for each work

   title, artist, year, medium, dimensions, edition info, inventory number

2. 2
   [ ] Multiple high-resolution images per record with the ability to mark a primary image
3. 3
   [ ] Provenance and ownership history as a proper field/log, not a free-text note
4. 4
   [ ] Exhibition and location history tied to each work
5. 5
   [ ] Export your full dataset and all original images at any time, in a standard format, without asking the vendor
6. 6
   [ ] Search and filter across every field (not just title/artist)

Tier 2 — Important, expect friction without it:

1. 1
   [ ] Condition reporting attached to the work, with date-stamped entries
2. 2
   [ ] Document attachments (COAs, invoices, consignment agreements, loan forms)
3. 3
   [ ] Rights and licensing fields (who holds copyright, license terms, reproduction restrictions)
4. 4
   [ ] Pricing and valuation history with dates
5. 5
   [ ] Role-based access so an intern doesn't see collector pricing
6. 6
   [ ] Reporting/report generation (insurance schedules, loan lists, valuation summaries)

Tier 3 — Genuinely useful, not deal-breakers:

1. 1
   [ ] Contact/collector linkage (which work is with whom)
2. 2
   [ ] Barcode or QR tagging for physical location tracking
3. 3
   [ ] Public-facing catalogue or viewing-room integration
4. 4
   [ ] Batch import/edit
5. 5
   [ ] Audit trail of who changed what and when

The most common mistake is over-weighting Tier 3 features — the public catalogue, the polished front-end — while under-weighting the boring Tier 1 stuff, especially full data export. If you can't get your data out cleanly, you don't own your archive. You're renting it.

## SaaS vs self-hosted: the trade-off that decides your cost model

This fork shapes your budget, your IT burden, and your risk profile. Most small galleries should land on SaaS, but it's worth making that choice deliberately rather than by default.

A two-to-five person gallery almost never has the technical bandwidth to self-host responsibly. "Self-hosted" sounds like control, but in practice it usually means one volunteer with some IT knowledge becomes the single point of failure. When they leave, security updates stop, and eighteen months later you're running an unpatched server holding your entire collection history.

SaaS shifts that burden to the vendor — which is why the export question from Tier 1 becomes your insurance policy. You accept vendor dependence in exchange for someone else handling backups and security, and your protection is the ability to leave with your data intact.

When self-hosting actually makes sense: you have a genuine in-house technical person (not a well-meaning volunteer), specific data-sovereignty requirements, or a collection large and specialized enough that off-the-shelf SaaS can't model it properly. For most artist-run spaces, none of those apply.

## The RFP questions that separate real vendors from brochures

You don't need a formal 40-page RFP. You need a tight list of questions that force vendors off their marketing script. Send these by email and pay attention to how they respond — vague answers on data ownership and backups are a red flag.

1. 1
   Can we export all records and all original-resolution images ourselves, any time? In what format?
2. 2
   If we cancel, how long do we have to retrieve data, and is there a fee?
3. 3
   Who legally owns the data we enter?
4. 4
   How often are backups taken, where are they stored, and how do you verify they can be restored?
5. 5
   What's your recovery process if a server fails — and what's the realistic recovery time?
6. 6
   Are images stored at original resolution, or compressed?
7. 7
   What access-control roles are available out of the box?
8. 8
   Is there an audit log of changes?
9. 9
   How is data encrypted, at rest and in transit?
10. 10
    How long have you been operating, and what happens to our data if the company shuts down?
11. 11
    What does onboarding/migration support include, and what does it cost?

That last continuity question matters more than people give it credit for. Small archive tools get acquired or quietly wind down more often than you'd expect, and galleries get 60 days' notice to move years of records. If a vendor gets awkward about the "what if you shut down" question, that's information.

## Metadata mapping and migration: where projects actually stall

The software choice isn't what kills these projects. The migration does. A gallery buys a good tool, then spends four months not moving anything into it because the existing data is a mess and nobody scoped the cleanup work ahead of time.

1. 1
   Inventory your current sources. List every place data currently lives — the master spreadsheet, the invoicing system, image folders, physical binders. You almost always find more sources than you expected.
2. 2
   Define your canonical fields. Decide the exact fields your new archive will use and their format — how you write dimensions, how you record edition numbers, date formats. Write it down as a one-page standard.
3. 3
   Build a mapping table. For each source, map its columns to your canonical fields. "Column F in the sales sheet = valuation, dated to the invoice date." This document prevents chaos during import.
4. 4
   Clean before you import, not after. Fix duplicates, standardize artist names, resolve conflicting titles in the source data. Importing garbage just relocates it.
5. 5
   Migrate a pilot batch first. Move 25–50 works, check them record by record, confirm the mapping held. Fix problems before doing the full set.
6. 6
   Full migration, then verify a sample. Import everything, then spot-check a random 10% against the source.
7. 7
   Attach images and documents. Link images to records — this is usually the most tedious step and the one most likely to get skipped.

The single biggest time sink is inconsistent artist names and titles across sources. "Untitled (2018)" vs "Untitled #4" vs "Untitled, 2018" for the same work will multiply your records if you don't standardize first. Budget real time for step 4.

“

Fix duplicates and standardize artist names before importing to avoid multiplying records.

Here's a quick visual of that migration workflow.

> Migration process at a glance: > Source audit → Field standardization → Mapping table → Data cleanup → Pilot import (25–50 works) → Full import → Sample verification → Image and document attachment

## Access control patterns for small teams

Small galleries tend to run wide open — everyone can see and edit everything — until the day an intern accidentally overwrites a valuation or a part-time assistant sees collector purchase prices they shouldn't. Access control isn't about distrust. It's about limiting the blast radius of ordinary mistakes.

1. 1
   Admin (1–2 people)

   full access, can change structure, manage users, delete records.

2. 2
   Registrar/full editor

   can edit all records and pricing, can't manage users or delete.

3. 3
   Staff/limited editor

   can add and edit works and condition notes, cannot see collector-facing pricing or valuations.

4. 4
   Read-only/intern

   can view and search, cannot edit anything.

If you're building out volunteer and intern involvement, this maps closely to how you'd structure any access system — and it pairs well with the thinking in the gallery accessibility system for small teams with KPI alignment approach, where you decide who touches what based on role, not on who happens to be around.

## A real scenario: what implementation actually looks like

The gallery: an artist-run space, three staff plus rotating volunteers, roughly 600 works across their own inventory and consigned pieces.

The problem: records lived in a shared spreadsheet with about 40 columns, plus image folders named inconsistently, plus a binder of physical COAs. Pulling together a loan list for a museum request took most of a day. Two works turned out to be recorded twice at different prices, which surfaced awkwardly during a sale.

What they did: chose a SaaS collection management tool after getting satisfactory answers to the export and backup questions. Spent about three weeks just on data cleanup before importing anything — mostly standardizing artist names and killing duplicates. Migrated a 40-work pilot, fixed the mapping, then ran the full import over a weekend.

Rough outcome: loan list generation dropped from most of a day to under an hour. Duplicate records went to zero after cleanup. The genuinely valuable part wasn't speed, though — it was that when their longtime registrar went on leave for two months, a volunteer could find and produce records without her. The institutional memory stopped living in one person's head.

Note what didn't happen: no dramatic revenue jump, no transformation. Archive software doesn't make you money directly. It stops specific, expensive failures — and it makes your operation survivable when a key person is out.

## A 6–12 week rollout plan with KPIs

Weeks 1–2 — Scope and select. Name your risk scenarios. Send RFP questions to 3–4 vendors. Do demos with your actual data questions in hand.

Weeks 3–4 — Decide and prepare. Choose the tool. Define canonical fields. Build the mapping table. Start data cleanup in the source.

Weeks 5–6 — Pilot migration. Move 25–50 works. Verify record by record. Fix mapping and field definitions.

Weeks 7–9 — Full migration. Import all records. Attach images and documents. Spot-check a 10% sample.

Weeks 10–11 — Access and training. Set up roles. Train staff and key volunteers on the workflows they'll actually use — adding a work, logging a condition note, pulling a report.

Week 12 and beyond — Adoption. Retire the old spreadsheet as the source of truth. Make the new system the only place records get updated.

KPIs worth tracking:

1. 1
   Time to produce a loan or insurance list (before vs after)
2. 2
   Percentage of works with complete Tier 1 records
3. 3
   Number of duplicate/conflicting records (target

   zero)

4. 4
   Percentage of works with at least one high-res image attached
5. 5
   Number of records updated outside the system after go-live (target

   zero — this measures real adoption)

That last KPI is the one everyone forgets. If people keep updating the old spreadsheet "just for now," you've bought software and kept your problem. Adoption is the whole game.

## Who should NOT buy archive software yet

A few honest cases where holding off is the right call.

If your collection is under about 50 works and stable, a well-structured spreadsheet with a disciplined naming convention and a real backup routine might genuinely be enough for now. Buying software to manage 40 works can be more overhead than it's worth.

If nobody on your team will own the migration and enforce that the new system is the source of truth, don't buy yet. Software with no owner becomes an expensive, half-populated ghost — and it makes things worse, because now your records are split across two places.

And if you haven't named the specific failure you're solving, wait. You'll pick the wrong tool.

## Bringing it together

The gallery archive decision gets treated as a technology purchase, but it's really an operational one. The tool matters less than whether you scoped the right risk, cleaned your data before importing it, set access so ordinary mistakes stay contained, and got your team to actually use the thing.

The same discipline that helps you turn visitor flow into curation decisions or keep press work from draining staff time during openings connects directly to this. Good records aren't an archive project sitting off to the side. They're the backbone that lets a small team answer any question — from a collector, an insurer, a lender, or a new hire — without the answer depending on one person's memory.

Buy against the failure you're trying to prevent. Protect your exit with clean data export. Scope the migration honestly. Measure adoption, not features. Do those four things and the specific tool you choose matters a lot less than the market makes it seem.

Buy against the failure you're trying to prevent. Protect your exit with clean data export. Scope the migration honestly. Measure adoption, not features. Do those four things and the specific tool you choose matters a lot less than the market makes it seem.