Deduptio

How to normalize company domains in Attio before matching

The short answer

Normalize in three passes, from safe to judgment. First, a string rule: lowercase, and strip the protocol, a leading www., the path and a trailing dot. That pass is safe to run on every record. Second, subdomains such as app.acme.com: collapse them to the root domain only when the root belongs to the company and not to a platform such as myshopify.com. Third, old and secondary domains: only an HTTP check shows that oldacme.io redirects to acme.com. Then match, and merge any pair whose clean domains agree.

This page is about the domain values themselves: getting https://www.Acme.com/pricing, acme.com and app.acme.com into a shape where a match rule can compare them. Choosing a match rule for companies as a whole, including the duplicates that share no domain at all, is in finding duplicate companies in Attio.

What Attio already does to a domain

Five facts from Attio's documentation decide how much of the work is left to you.

Two things are not documented. Attio does not say whether it strips the protocol or a www. prefix, and it does not say whether company matching and duplicate detection compare the full domain or the root domain. Treat both as unknown. After your next import, open five records whose source data had https://www. in it and read what landed under Domains.

Per Attio's domain attribute type, Companies object and merge and delete records pages, checked November 10, 2026.

Five ways one company ends up on different domain strings

Each variant below needs a different kind of fix. The first two are string problems; the last three are questions about the company.

VariantExampleWhat resolves it
Protocol, www, pathhttps://www.acme.com/ vs acme.comString rule, safe on every record
Case and trailing dotACME.com. vs acme.comString rule, safe on every record
Subdomain of the company's own domainapp.acme.com, eu.acme.com, mail.acme.comRoot domain, after checking who owns the root
Old or secondary domain that redirectsoldacme.io → acme.comAn HTTP check; no string rule can see it
Country or product domain with its own siteacme.de, getacmeapp.comA decision; see several domains or subsidiaries

Pass 1: strip protocol, www and path

This pass changes the spelling of a domain and never its meaning, so it is safe to apply to every value in the workspace. Apply the steps in this order:

  1. Trim spaces and lowercase the value.
  2. Remove the scheme: https://, http:// or anything else ending in ://.
  3. Remove a leading www.. Only at the start, and only that exact label: www2.acme.com and wwwacme.com are left alone.
  4. Cut everything from the first /, ? or #.
  5. Remove a trailing dot: acme.com. is a valid fully qualified name for acme.com.

After this pass, https://www.Acme.com/pricing?ref=ad and acme.com are the same string. The order matters: strip the scheme before www., or https://www.acme.com keeps its www. because the string does not start with it.

The pass does not touch other subdomains. app.acme.com stays app.acme.com, and that is deliberate: whether a subdomain is the same company is the next question, not a spelling question.

Pass 2: decide what to do with subdomains

Most subdomains on a company record are the company's own: app. for the product, eu. for a region, mail. or go. for marketing. Collapsing these to the root domain is right, and it is how you find the record created from jan@mail.acme.com next to the one on acme.com.

The root domain is not always the company, though. Collapse blindly and these cases go wrong:

A practical rule: collapse a subdomain when the root domain has its own record in your workspace, or when the root domain's website is plainly the same company. Otherwise keep the full domain.

To see every subdomain at once, read root_domain through the API. Query the Companies object with POST /v2/objects/companies/records/query, take each value of domains, and group records by root_domain. Any root with more than one record is a candidate pair. The request shape is in duplicate detection with the API.

Pass 3: resolve redirects

A rebrand or an acquisition leaves an old domain that redirects to the new one. The two strings share nothing, so no string rule can connect oldacme.io to acme.com. You have to ask the web.

Open the old domain in a browser and read the address bar after it loads. For a list of domains, run this in a terminal and read the final address:

curl -sIL -o /dev/null -w '%{url_effective}\n' https://oldacme.io

Read the result in one of four ways:

A redirect shows where a domain points today, not where it pointed when the record was created. A domain that changed owners redirects to the new owner. Treat a redirect as evidence for a merge, not as the merge decision.

Write the clean domains back to Attio

Normalizing in a spreadsheet changes nothing in Attio. The values that matter are the ones under Domains on each record, because those are what email sync matches new people against.

  1. Open the company record and edit Domains: add the clean root domain and keep any subdomain your contacts actually email from.
  2. If Attio refuses the clean domain, another record already holds it. That record is the duplicate. Merge the two instead of editing.
  3. To merge, click the ⋮ icon in the top right of one record, choose Merge records, add the other record and put the record whose values you trust on the right. The full steps and the survivor choice are in how to merge records in Attio.
  4. Open the merged record and check that every domain from both originals is under Domains. Attio does not document how a merge combines multi-value attributes. Add any domain that dropped off: the record that held it is gone, so Attio now accepts it.

Keep the old and variant domains on the merged record rather than deleting them. A domain that leaves the record is a domain the next email can create a new company from. A merge in Attio has no documented undo, so check the pair before you click.

A Website text field is not a domain

Imports often bring a Websitecolumn with full URLs in it. Mapped to a text attribute, it never goes through Attio's domain handling, and every tool that compares text compares it as text. A spreadsheet sees https://www.acme.com/ and acme.com as two values.

Run the URLs through pass 1 and put the result into Domains. Then match on Domains, not on the text field. The text field can stay as a record of what the source system said.

What Deduptio normalizes, and what it does not

Deduptio applies pass 1 for you. When a match rule keys on an attribute of Attio's domain type, such as Domains on Companies, every value is lowercased and stripped of the scheme, a leading www., the path, query and fragment, and a trailing dot before comparison. Every value of a multi-value attribute is compared, so two records match if they share any one domain after normalization.

The limits, stated plainly:

Each group shows why its members matched, and you decide per group before anything merges. Merges run through Attio's native merge. Details are in match rules and plans.

Match companies on clean domains

Connect your Attio workspace, write one match rule, and run a scan. Scans are read-only — you see every group and the reason it matched before anything is merged.

Start a free Attio duplicate scan

Checklist

  1. Open five recently imported companies and check whether www. or a protocol landed in Domains.
  2. Run pass 1 on every domain and every Website URL: case, scheme, www, path, trailing dot.
  3. Group by root domain and decide each subdomain group: company subdomain or platform.
  4. Check old and secondary domains for redirects; treat a redirect as evidence, not proof.
  5. Write clean domains into Domains; merge where Attio says the domain is taken.
  6. After each merge, confirm every domain from both records is still on the result.

Related