Explore our new Cybersecurity Terms Glossary, explained by our experts.
Cybersecurity Terms Glossary
Normalisation maps every source into one field vocabulary so a single query can span them. Enrichment adds the context the raw record lacks. Without both, cross-source detection is guesswork.

What Are Log Normalisation and Enrichment?
Thirty log sources describe the same concepts in thirty ways. One calls the originating address src_ip, another source.address, a third ClientIP. One records time in local time without an offset, another in epoch milliseconds. Normalisation is the work of mapping all of them onto a single vocabulary so that one query can ask a question of all of them. Enrichment is the separate work of adding what the record never contained.
Do not invent your own. Four established schemas cover most situations: the Open Cybersecurity Schema Framework, launched in August 2022 with backing from a large vendor group and now the most vendor-neutral option; Elastic Common Schema, widely implemented and the basis of OCSF’s event model in part; the Common Information Model in Splunk; and the Advanced Security Information Model in Microsoft Sentinel. Pick the one your primary platform supports natively and resist local variations, because every deviation becomes a permanent translation cost.
Normalisation covers more than renaming fields. Timestamps need converting to a single timezone with explicit offsets, and clock skew between sources needs monitoring, since a drifting host silently disables every time-window detection that involves it. Values need standardising too: one firewall says deny, another DROP, a third returns a numeric code.
The hardest part is identity. The same person appears as jsmith in Active Directory, DOMAIN\jsmith in Windows events, [email protected] in the cloud identity provider, [email protected] in Kerberos, and an opaque identifier in a SaaS application. The same host appears as a short name, a fully qualified name, several addresses over the course of a week, and a cloud instance identifier.
Until these are resolved to a canonical entity, cross-source correlation cannot work, risk-based alerting cannot accumulate score against the right user, and investigations consume hours on manual stitching. Solving it properly means maintaining mapping tables from the directory and from DHCP or IP address management, applying them at ingestion, and monitoring the proportion of events that fail to resolve. That last metric is a good proxy for the health of the whole pipeline.
Enrichment at ingestion is fast to query but freezes a snapshot: an event enriched when the user was a contractor still says contractor after they become an employee. Enrichment at query time is always current but slow and dependent on the reference source being available. The usual compromise is to enrich stable attributes at ingestion, keep volatile ones for query time, and timestamp the reference data so an investigator knows what was true when.
Normalised and enriched records are larger than raw ones, sometimes two or three times. On volume-priced platforms that is a real budget line, and the common answer is to normalise the fields detections need rather than every field present. Keep the raw record as well, because parsers have bugs and an investigation occasionally needs the original bytes.
Parser breakage is the recurring operational problem. A vendor changes a format in a minor upgrade, the parser quietly maps nothing to the field a rule depends on, and the detection stops firing without any error. Monitor the proportion of events failing to parse per source, alert on change, and re-run detection tests after every platform upgrade.
Finally, enrichment adds personal data to records that may not have contained it. Joining employee attributes onto security events creates a richer profile than either source alone, so the retention and access controls described under log management apply to the enriched store with more force, and the processing belongs in your documentation under UK GDPR.
Choose the one your primary analytics platform supports natively, because fighting the platform costs more than the theoretical benefits of a different schema. OCSF is the most vendor-neutral and is gaining adoption; Elastic Common Schema is mature and widely implemented; Splunk and Microsoft Sentinel each have their own established models. Whatever you pick, document local extensions explicitly rather than allowing informal variations to accumulate.
Both, split by how volatile the attribute is. Enrich stable attributes such as asset criticality and network segment at ingestion, so every event carries them and queries stay fast. Leave volatile attributes such as current risk score or live threat intelligence verdicts to query time. Timestamp the reference data used, so an investigator can tell what was true at the moment of the event.
Because almost every multi-source detection depends on recognising that two records describe the same user or host. Without consistent resolution, correlation rules match nothing, risk scores accumulate against fragments of an identity, and investigators stitch timelines by hand. Maintain mapping tables from the directory and address management systems, apply them at ingestion, and monitor the share of events that fail to resolve.
Detections that depend on the affected fields stop firing, usually without any error being raised, because an empty field simply fails to match. The causes are normally vendor format changes during upgrades. Monitor parse failure rates per source, alert on sudden change, keep the raw record alongside the parsed version, and re-run detection tests after every platform or agent upgrade.
You May Be Interested In These
We use a few cookies that are needed to make this website work, and they are always on. We would also like to set analytics cookies, to see which pages are useful; functional cookies, so that our live chat works; and advertising cookies, to measure our campaigns. We will not set any of those until you say yes, and you can change your mind at any time from the Cookie Settings link in the footer.
Choose the cookies you are happy for us to use. Strictly necessary cookies are always on; everything else is off until you turn it on.
These cookies keep the website working: they hold your security checks, your form submissions and the choice you make here. The site cannot run without them, so they are always on.
These let our live chat open and remember your conversation, and let the newsletter sign-up work. Switch them off and the chat window will not appear.
These tell us which pages are read and how people find us, so that we can improve the site. The reports are about groups of visitors, never about you personally.
These let us measure our advertising and show you our adverts on other websites, such as LinkedIn, Google and Meta. Switch them off and you will still see adverts, but they will be less relevant.