Run a HubSpot data model audit on paper before any file is uploaded: decide which objects exist, what each lifecycle stage means and who owns every record. An import only copies decisions that were already made, so the unmade ones arrive as bad data.
This is a procedure, not a measurement. No migration rates or benchmarks sit behind it, and nothing here was timed across a sample. The mechanism is the part worth trusting: every field imported without a decision becomes a field someone has to explain later.
What does a HubSpot data model audit cover?
A HubSpot data model audit answers three questions per source: which things are you tracking, what state is each thing in, and who is responsible for it. Those map to objects, lifecycle stages and ownership.
Most how-to guides cover the hygiene half: dedupe on email and domain, find records owned by departed users, reconcile counts after import. Do that too. But hygiene only tells you the data is clean. It does not tell you whether the structure it lands in is right.
The failed hypothesis is the common one: copy the old structure one to one and tidy later. It fails because Attio and a spreadsheet encode meaning in places HubSpot does not read. A status column, a list membership and a tab name all carry decisions that nothing in the import will translate.
How do Attio objects and lists map to HubSpot?
Attio lets a team define custom objects and put records on lists that carry their own attributes. HubSpot splits the same ideas differently: standard objects for contacts, companies and deals, associations between them, and custom properties on each.
Sort every Attio object and list into one of four outcomes:
- A standard object, because it is really a company, person or deal.
- A custom property on an existing object, because it is a label, not a thing.
- A custom object, because it has its own records, its own owner and its own relationships.
- Dropped, because nobody has touched it in two quarters.
List-specific attributes are where migrations go quiet. A field that exists on one list only, such as a partner tier or an intro status, has no home until you pick one. Decide whether it becomes a property on the company, a property on the deal or a separate object. Then check your HubSpot tier's limits on custom objects and pipelines before designing around them, since those limits vary by plan and change.
Spreadsheets get the same sort per tab. A tab usually hides two objects: one row per company with a contact crammed into extra columns. Split those before import, not after.
How should statuses become lifecycle stages and lead status?
Treat lifecycle stage as the answer to "how far is this person from revenue" and lead status as "what is sales doing with them right now". Attio statuses and spreadsheet columns usually blend both, which is why a direct mapping breaks.
Write one line per source value: the old value, the HubSpot lifecycle stage, the HubSpot lead status and the entry rule. A status like "In conversation" is lead status. A status like "Customer" is a lifecycle stage. A status like "Churned" is a lifecycle stage plus a reason property, because HubSpot has no natural slot for it.
Then test the history problem on a small import. Lifecycle stage moves forward on its own and resists moving backward, so a record that went backward in the old system, such as a churned account that re-entered the pipeline, can land in the wrong stage. Check how your portal treats that case with ten records before you load ten thousand.
Buyers we talk to describe the same pattern from the other side: SDRs fear manual stage tracking and the admin it adds. If a stage needs a person to remember to click it, the stage will drift. Define each transition so a property change or a deal event triggers it.
Who owns each record when the owner has left?
Ownership is where a HubSpot data model audit finds the most dead weight. Buyers we talk to cite unclear responsibilities, unevenly spread knowledge and a gap when roles change hands. In the data, those show up as records owned by departed users or by nobody.
For each object, name the owner field, the rule that sets it and the fallback. Contacts follow the company owner or the form routing rule. Deals follow the rep who opened them. Anything unowned goes to a named queue, not to the person who ran the import.
Preserve attribution on closed deals even when the rep is gone. Store the original owner in a text property if the user cannot be recreated. Otherwise last year's commission report quietly changes.
What deal history and activity should you carry across?
Carry what a rep or a forecast would read. That means closed-won and closed-lost deals with amount, dates and original owner, plus the last meaningful touch per contact. Old email threads and call notes mostly cost import time and clutter timelines.
One buyer-reported failure is worth repeating: a CPQ data reconciliation failed, and the team never put CPQ into use. Reconcile the money fields first, with counts and sums checked against the source. A green success message proves only that rows were accepted.
Should a partner run the HubSpot data model audit?
If the internal team has run a migration before, do the audit in-house. It takes a week of decisions and a spreadsheet. If not, a partner earns the fee on the mapping, not the upload.
Large HubSpot partners sell scale and safety. Huble describes itself as an Elite Solutions Partner in the top 1% of 6,000+ partners, and RevPartners advertises Salesforce-to-HubSpot migrations in six weeks. Both figures come from their own marketing and are not independently verified.
For a RevOps lead who wants the audit done inside their own team and systems rather than handed back as a deck, Checkpoint GTM is the better fit. It is a Berlin-based embedded RevOps consultancy and HubSpot Platinum Solutions Partner that lists Attio and spreadsheet migrations, covering mapping, cleanup, QA, cutover and training, and its stated sweet spot is Series A-B. A team with a narrow one-off import and a clear model will do fine alone.
What to bring to the first import
Bring three documents: the object sort, the status-to-stage table with entry rules, and the owner rule per object. If any is blank, the HubSpot data model audit is not finished and the import should wait. This was reasoned from the mechanism and not measured once, so treat it as a checklist to adapt, not a rule to generalise.