Why contacts need the right company
A contact without a company is a person with no context. The seller who calls them does not know their employer is already a customer, the scoring model reads fit from nothing, and the pipeline report cannot say which accounts the marketing reached. A contact attached to the wrong company is worse, because the context is confidently wrong: a trial user from one firm routed to the seller who owns a different firm with a similar name, or a new buyer at a customer treated as a stranger because their sign-up created a second record for the same organisation.
The cost compounds through every system that reads the company. Lead routingRules that assign each new lead to an owner, queue or next step based on territory, segment, score or source. sends people to the owner of the company they are attached to. The lead stages and pipeline guide counts opportunities and conversions that sit on company records. Account-basedA B2B approach that targets a chosen list of companies with coordinated marketing and sales. programmes, as the account-based marketing guide describes them, work a named list of organisations and measure engagement at account level, which only means something if the people from each organisation are attached to its record. And the ideal customer profileA description of the type of organisation most likely to buy, succeed with the product and stay, used to choose target accounts., as the tutorial on building one from CRM data sets it out, is built from the customers the business already has, read from company records whose contacts and deals are complete.
Matching is therefore not a hygiene task to be done once, but a set of rules applied every time a contact is created: at the form, at the import, at the integration and when a seller adds someone by hand. The rules below are general practice rather than any CRM's behaviour; many CRMs offer some automatic association, and the point of writing the rules down is that the team knows what should happen whether the system does it or a person does.
Matching on the email domain
The first rule is the simplest and covers most contacts: a work email address carries the company's domain, and the domain identifies the company. The fundamentals guide prevents duplicate companies by matching on website domain at every entry point, and the same key attaches contacts. A contact whose address ends in a domain that belongs to an existing company record is attached to that record; a contact whose domain matches no record either creates a new company with that domain or waits for review, by a rule the team chooses in advance.
The domain is enough when three things are true: the address is a work address, the domain belongs to one organisation, and that organisation has one company record. Each condition fails often enough to need its own rule. The domain stored on the company record also needs care: it should be the bare domain, without the "www" prefix or a path, held in a field that is checked for duplicates, because two company records with the same domain are the duplicates that matching is meant to prevent. DeduplicationRemoving double counts when one conversion is reported by more than one source. of companies comes before matching contacts to them; matching contacts into a set of duplicate companies just spreads them more evenly. When duplicate contacts keep returning after every merge, the diagnosis for duplicate contacts in the CRM separates genuine two-address pairs from the entry points that fail to match.
| Situation | Is the domain enough? | Rule as general practice |
|---|---|---|
| Work address, one domain, one company record | Yes | Attach to the record with that domain |
| Work address, domain matches no record | Partly | Create the company or queue for review, by an agreed rule |
| Personal address from a consumer mail service | No | Leave unmatched; review using what the person stated |
| Shared domain used by several organisations | No | Match on the company stated and reviewed, not the domain |
| Company with several domains | Only with a domain list | Hold every known domain on the one record |
| Subsidiary or regional office with its own domain | Yes, to the subsidiary | Attach to the subsidiary's record, which links to its parent |
| Subsidiary or office sharing the parent's domain | Not to the right level | Attach to the parent, then assign the unit by review or territory |
Personal addresses, shared domains and companies with several domains
Personal addresses. People sign up for trials, webinars and downloads with personal addresses from consumer mail services. The domain says nothing about their employer, and a rule that creates a company for the mail service, or guesses from a similar name, attaches people to the wrong place. The safer rule is that these contacts stay unmatched until there is evidence: the company name and website the person gave on the form, confirmed by a reviewer; a later sign-up from a work address that can be merged with the first; or a seller's conversation. While they wait, they need an owner and a route, so a review queue with a named owner is part of the design rather than a backlog. Forms can help by asking for a company name where the next step genuinely needs one, weighed against the completions every extra field costs. The same review applies when a customer's champion moves to another employer and writes from a new address, and the guide to a customer champion changing jobs keeps the old record as history while the new contact is checked and attached to the right company.
Shared domains. Some domains are used by many organisations: a university's domain covering separate departments that buy independently, a group-wide domain covering companies that are separate customers, or a domain belonging to an agency that buys on behalf of clients. Matching on the domain alone puts all of them on one record. Where the business knows a domain is shared, list it, and send its contacts to review so they can be attached on what they stated rather than on the address.
Several domains. Organisations rebrand, acquire other firms and run regional or product sites, so one company can send people from several domains. The fix is a list of domains on the one company record, with the primary website kept separately, so that each additional domain matches the right record instead of creating a new one. When a review finds a second domain, adding it to the record is part of resolving the case.
Parents, subsidiaries and regional offices
Groups of companies need a structure, and the choice is not whether to have one record or many, but how to join the many. The common pattern is a company record for each unit that buys or decides on its own, a subsidiary, a division with its own budget, a regional office that signs its own contracts, joined to a parent record that represents the group. Contacts attach to the unit they work for. Reports can then answer two kinds of question: about one unit, by reading its record, and about the group, by rolling the units up through the parent.
The test for whether a unit deserves its own record is commercial, not legal: does it buy separately, have its own decision-makers or have a different owner in the sales team? If not, it can stay inside the parent's record with a field for the site or region. The expansion and renewal guide maps the parts of a customer that do not use the product yet, such as new teams, sites and subsidiaries, and that map is far easier to keep when each part has a record the map can point to.
The structure needs an owner of its own. Someone decides when a new unit gets a record, links it to its parent and records the domains it uses. Without that, groups end up flat in some places and nested in others, and a roll-up report quietly misses the units nobody linked.
Ownership and routing once a match is made
A match decides more than where a contact is stored; it usually decides who calls them. The playbook for inbound lead handling routes leads by rules that include existing account ownership: a lead from a company that already has an owner goes to that owner, with a fallback if the owner is away. The matching rules and the routing rules therefore have to be designed together. A contact attached to a customer record should reach the account owner, not the next seller in the rotation; a contact attached to a target account in an account-based programme should reach the seller named for that account.
Some exceptions are worth writing down rather than leaving to judgement on the day. A request for a product the account owner does not sell may go to the seller who does, with the account owner told. A contact from a subsidiary with no owner of its own may go to the parent's owner. A contact still waiting in the personal address queue may be routed by territory until the review is done. Each exception is a rule in the routing design, and each one is a case to include in the test leads of the lead routing QA checklist, which checks that leads are deduplicated against existing contacts and accounts before they are assigned. When live leads from known accounts still reach the wrong seller, the diagnosis for leads assigned to the wrong salesperson separates a wrong match from rules read in the wrong order, stale rotation groups and duplicates.
Account-level scoring and reporting
Matching changes what the scoring model sees. The lead scoring guide reads fit attributes such as company size and industry from the company record, alongside what the form and other sources provide. A contact matched to the right company inherits the right fit; a contact matched to nothing or to the wrong company is scored on a guess. Where the business also looks at engagement across all the people from one organisation, as account-based programmes do, matching decides whose activity is added together.
Reporting changes the same way. The B2B measurement guide records sourced pipeline as the first touch on the account and influenced pipeline as any touch on the opportunity. Both need contacts attached to the company that owns the opportunity: a webinar attended by three people from one customer group, recorded on three unconnected company records, credits nobody with the deal their parent signs. Read through the marketing funnel framework, matching is what lets the funnel be counted in accounts at the stages where accounts are the unit of decision, while the earlier stages are still counted in people.
When matching rules change, the reports change with them. Re-matching a backlog of contacts can move historic pipeline from one account, or from no account, to another. Record the change and its date where the team keeps its measurement changes, so a step in an account-level trend is read as the rule changing, not the market.
Zoravique Payroll fixes its matching
Zoravique Payroll, an invented payroll software firm, sells to employers with several legal entities. A quarterly review finds that trial sign-ups from one retail group sit on 5 separate company records, created from 3 domains and two misspelled company names, and were routed to 4 different sellers. Across the database, 1,860 of 7,400 trial contacts in the last half year signed up with personal addresses and were attached to nothing. The team makes three changes. Domain rules: each company record holds a list of known domains, and the 3 domains of the retail group now point to one record. A parent and subsidiary structure: the group becomes a parent with 6 subsidiary records, one per employing entity, each with its own owner where it buys separately. And a review queue for personal addresses: contacts are matched from the company name they gave, confirmed by a named reviewer, and routed by territory while they wait. The next quarter, sign-ups from existing groups reach the account owner, and the retail group's trials, meetings and opportunity appear on one roll-up for the first time.
The example is specific, but the order generalises: deduplicate companies, hold every domain on the right record, build the group structure, route personal addresses through review, then rebuild the reports.
Common mistakes
- Matching contacts into duplicate companies. Two records with one domain split everything that follows. Deduplicate companies first.
- Guessing from personal addresses. A similar name is not a match. Keep personal-address contacts in a review queue with an owner.
- Treating a shared domain as one customer. Departments, group companies and agencies that buy separately end up on one record. List shared domains and review them.
- One domain per company. Rebrands and acquisitions create new records for existing customers. Hold every known domain on the record.
- A flat list for a group of companies. Without parents, the group cannot be rolled up; without subsidiaries, nobody can see which part is buying.
- Routing by rotation after a match. A customer's new colleague reaches a seller who has never heard of the account. Route matched contacts to the account owner, with written exceptions.
- Changing matching rules without noting it. Historic pipeline moves between accounts and the trend line breaks for no visible reason.
