Every Data 360 project eventually comes down to one question: is this the same customer?
Your Sales Cloud contact, your Service Cloud case, your Marketing Cloud subscriber and your ecommerce account might all be one person. Or they might be four. Identity resolution is how Data 360 decides, and almost everything you build afterwards depends on getting that answer right.
Segments, calculated insights, service views and Agentforce all work from the unified profile. If the profile is right, they shine. If it is wrong, every one of them inherits the problem.
What identity resolution actually does
If you knew Data 360 by its earlier name, Data Cloud, identity resolution works the same way. It runs on a ruleset with two parts:
Match rules decide which source records belong to the same individual. Each rule compares fields, such as name, email, phone, address or a party identifier like a loyalty or customer number, using a match method. Depending on the field, that can be exact, normalised (ignoring formatting differences, useful for email, phone and address) or fuzzy (tolerating small spelling differences).
Reconciliation rules decide which value wins when matched records disagree. For example, take the value from the most recently updated record, the value that appears most often, or the value from the source you trust most.
The result is a unified individual: one profile linked back to every source record that contributed to it. Your source records are not changed. Unified link objects act as the bridge between them and the profile.
That design is a gift. It means you can tune your rules and rerun them without touching the data your teams work in every day.
Start with the sources you trust
Unified profiles are only as trustworthy as the data that feeds them. Before writing a single rule, ask two questions of every source you plan to connect:
Which identifiers are reliable here? A verified customer number from your billing system is worth more than an email typed into a web form.
Who owns the quality of this data? If nobody does, start there.
A small number of sources people already rely on is a far better starting point than connecting everything at once. You can always add more. Fixing data before it lands in Data 360 is easier than cleaning it up afterwards.
Choosing how strict to be
Every match rule is a trade-off between catching real duplicates and avoiding false merges. Salesforce frames it as choosing between high consolidation and a more conservative approach.
Some practical guidance:
Lead with your strongest identifiers. If you have a reliable identifier, such as a customer or member number, a rule matching on that alone is often your most dependable.
Combine weaker signals. A first name on its own proves very little. A fuzzy name match plus a normalised email, or a name plus a normalised address, is much more convincing.
Use fuzzy matching where typos really happen, like names, and keep it away from identifiers that should match exactly.
Add rules one at a time. More rules usually means a higher consolidation rate, but you want to know which rule caused which merge.
Decide which source wins
Reconciliation is where a lot of quiet problems start. If your CRM says a customer lives in Parramatta and your ecommerce platform says Newcastle, which one does the unified profile show?
There is no universal answer, but there is a good process:
Set source priority for the fields where one system is the source of truth. Billing for addresses, perhaps, or your core CRM for names.
Use most recent for fields that genuinely change, like preferred contact details, where the latest update is usually right.
Be deliberate with IDs. When several sources hold similar identifiers, prioritise them by source so IDs stay consistent across profiles.
Write these choices down. When a service agent asks "why does the profile say this?", you want an answer that traces back to a rule and an owner.
Test on customers you know
This is the step most teams skip, and the one we would never skip.
Before you build segments or switch on activation, pick a test set of real people whose records you understand well:
Customers you know have accounts in several systems.
Customers with common names.
Staff who are also customers.
Households that share contact details.
Run the ruleset and check each one in Profile Explorer. Did the right records come together? Did anything merge that should not have?
Then look at the processing history for the ruleset, including the consolidation rate. A rate that is far higher or lower than you expect is a prompt to look closer. Data 360 also lets you build more than one ruleset, which makes it possible to compare a cautious set of rules against a more aggressive one before you commit.
Watch for merged strangers
Under-matching is annoying: two profiles for one person. Over-matching is worse: one profile for two people. That is when one person's offer, case history or consent lands on someone else's profile.
The usual causes are shared contact details:
Shared family emails, such as a household inbox used by several people.
Shared phone numbers, like a landline, a business switchboard or a parent's mobile on a child's account.
Placeholder values, such as "noemail@" addresses or dummy phone numbers entered to get past a required field.
If any of these exist in your data, do not let them carry a match on their own. Pair them with a stronger identifier, or clean the placeholders out at the source. A quick look at the most common email addresses and phone numbers in each source usually tells you where the risk is.
Keep it healthy as new sources join
Identity resolution is not set and forget. Every new source brings new identifiers, new formats and new quirks.
Each time a source joins:
Check its field mappings before the ruleset runs again, and review any warnings Data 360 shows.
Rerun your test set of known customers.
Compare the consolidation rate before and after.
Give the ruleset an owner, the same way you would own a critical integration. If you also need to keep processing costs under control, Data 360 lets you decide whether identity resolution jobs run automatically, so you can schedule runs around when your data actually changes.
Why it matters more every month
As more teams use Data 360 for real-time engagement and Agentforce, the unified profile stops being a reporting asset and becomes the thing your customer experience runs on. An agent can only be as helpful as the profile it reads.
Getting identity resolution right is not glamorous work. It is the foundation that makes everything Data 360 can do actually land, and it is very achievable with a clear process and a few honest test cases.
If you would like a second pair of eyes on your match and reconciliation rules, or help setting them up so your team can run them confidently, talk to us. We build so your team can run it.
#sustainablestacks