How To Segment By Website Behaviour

Website behaviour can help turn a broad email audience into smaller groups with a clearer reason to hear from you. Someone who has compared implementation pages, downloaded a technical guide or started a renewal form is unlikely to need the same message as a first-time visitor who viewed a single blog post.

The important distinction is between observing activity and using it well. A page view is not automatically intent. Nor does a useful-looking audience in an analytics platform automatically become an appropriate email segment. The best programmes collect a small number of meaningful signals, attach them to known people only where the connection is valid, and use those signals alongside consent, customer status and contact-pressure rules.

This article focuses on permission-based marketing and lifecycle email. It is not a method for identifying anonymous visitors and adding them to email marketing, or for pursuing people who have not asked to hear from you.

What website-behaviour segmentation means

Website-behaviour segmentation groups subscribed contacts according to actions they take on your site. Depending on the business and its systems, those actions may include:

• viewing a product, service or pricing page;
• using a calculator, product finder or comparison tool;
• starting, completing or abandoning a form;
• downloading a guide or registering for an event;
• adding an item to basket, beginning checkout or purchasing;
• logging in, using a feature or reaching an account-support page; and
• returning repeatedly to a defined area of the site.

In practice, a segment is usually a rule made up of four parts: who the person is, what they did, when they did it and what should stop the rule applying. For example: “Subscribed trial users who viewed the integrations page at least twice in the past 14 days, have not booked a demo, and have received fewer than two promotional emails in the past seven days.”

That is materially more useful than “people interested in integrations”. It makes the event, time window, audience and exclusions visible. Those details make a segment reviewable, testable and less likely to send a plainly irrelevant message.

Start with the email decision, rather than the data you can collect

Tracking plans often become a long list of clicks because modern tools make events easy to create. That creates noise, maintenance work and privacy risk without necessarily improving email.

Instead, begin with a decision you could make differently. Ask:

• What email, content or service message would change because of this behaviour?
• Would the recipient reasonably expect us to use this action in that way?
• Does the event indicate a need or stage that is stronger than a casual visit?
• What is the appropriate time window before the signal becomes stale?
• What subsequent event means the message is no longer helpful?

If there is no clear answer, do not create an email segment simply because the event exists.

For instance, a visitor who downloads an introductory guide may warrant a short educational sequence, provided they have asked for marketing and the form made that follow-up clear. A customer who has started a support request should generally be routed into service handling, not a promotional sequence. A person who reads a careers page may be exploring employment, rather than indicating buying intent. The same page view can mean different things in different contexts.

Build an event model that records context

A robust tracking set-up records more than a generic “page viewed” event. It should retain enough context to make later segmentation accurate, while avoiding unnecessary personal data.

For a product-view event, useful fields might include the product family, content category, account status, event time and a stable pseudonymous visitor or customer identifier. For a form event, the important distinction is usually the form’s purpose and outcome: viewed, started, validation error, submitted, confirmed or cancelled.

Avoid using a full URL as the main segmentation field where it can contain names, email addresses, order references, search terms or other information placed in query parameters. Create a controlled page or content taxonomy instead. For example:

Raw activity	Safer, more useful event label	Possible email use
Viewed several URLs in a product section	interest_area = reporting	Send relevant educational material, subject to permissions and frequency rules
Started a consultation form	form_started = consultation	Consider a short, helpful reminder only if the form was not submitted
Submitted a webinar form	registration_complete = webinar_2026	Confirmation, joining instructions and relevant follow-up
Visited the billing help centre	support_topic = billing	Usually exclude from unrelated promotional pressure; offer help where appropriate

Use controlled values wherever possible. A finite list of products, topics and form types is easier to audit than thousands of individual URLs. It also reduces the chance that a campaign unexpectedly uses sensitive or misleading information.

Track outcomes, not just starts

Form starts are often valuable because they reveal friction or unfinished intent. They are also easy to misread: a person may have opened a form by accident, encountered an accessibility issue or simply decided the offer was not right for them.

Record the full lifecycle. A simple model might look like this:

1. Form viewed: low-confidence interest.
2. Form started: stronger interest, but no commitment.
3. Form submitted: a completed request or registration.
4. Form confirmed: where a confirmation step is required, this is the authoritative outcome.
5. Downstream resolution: booked, purchased, qualified, declined, cancelled or supported.

The later events should suppress earlier-stage messages. A contact who has booked a consultation must leave the consultation-reminder audience immediately. A person who completes a purchase should not receive basket-abandonment prompts for that purchase. These exclusions are as important as the trigger itself.

Resolve identity carefully

Most website visits are anonymous. They can become linked to a person when someone logs in, submits a form, follows a tagged email link, or otherwise identifies themselves within a consented and properly disclosed journey. The method matters.

Do not treat an anonymous device identifier as a licence to email an individual. The email platform needs a reliable, permissible association between the website event and a contact record, and that record must be eligible for the proposed message.

Identity resolution also needs practical safeguards:

• Use a stable internal contact or account ID rather than repeatedly passing email addresses through URLs.
• Apply a defined merge rule where several devices or records appear to belong to one person.
• Keep source and timestamp fields, so a team can see whether an interest is current, historical or inferred.
• Test shared-device and shared-inbox edge cases, especially in household, education and business settings.
• Do not pass sensitive form content into advertising, analytics or email tools unless this has been assessed and is genuinely necessary.

Data-protection principles still apply to behavioural data. The ICO says personal data should be adequate, relevant and limited to what is necessary for the stated purpose, and organisations should be able to demonstrate that they have applied this principle. The ICO’s guidance on data minimisation is a useful test for every property retained in an event stream.

Make consent, transparency and preferences part of the segment

For UK organisations, using website behaviour to tailor marketing is profiling for direct-marketing purposes. It should be fair, transparent and consistent with the person’s permissions. The ICO notes that direct marketing includes activities that lead up to or support the sending of marketing, including targeting and profiling. Its guidance on identifying direct marketing explains why the preparation work matters as much as the send itself.

Website storage and access technologies used for behavioural tracking will commonly require consent under PECR. Consent needs to be meaningful: non-essential tracking should not be set before consent, and choices must be easy to make and withdraw. The ICO specifically advises particular care for behavioural-tracking technologies. See the ICO’s cookies and similar technologies guidance.

The email rule is separate but connected. Before a behavioural segment is activated, test eligibility for the email itself: marketing permission or another applicable basis, channel preference, regional requirements, existing customer relationship where relevant, and any suppression or objection. People have an absolute right to object to direct marketing, including related profiling, and objections must be honoured. The ICO’s guidance on collecting data and profiling also warns that profiles must be fair and transparent, particularly where they rely on more intrusive information or assumptions.

A useful operating approach is to make privacy controls available at three levels:

• Cookie and tracking choice: whether non-essential website tracking can occur.
• Marketing channel choice: whether the person can receive marketing by email.
• Topic and frequency choice: whether they want particular content or lower-volume communications.

Do not treat acceptance of website analytics as blanket permission for email marketing. Likewise, a marketing subscriber may be entitled to receive your newsletter without expecting every click and page visit to create an increasingly detailed profile. Explain the use plainly in your privacy information and at relevant collection points. Where the tracking or profiling is extensive, involves sensitive information, combines sources or could meaningfully affect people, obtain appropriate privacy advice and consider whether a data protection impact assessment is required. The ICO recommends DPIAs for several higher-risk cases, including large-scale profiling and matching data from multiple sources. Its DPIA guidance sets out those triggers.

Choose signals by strength and recency

Behavioural signals have different levels of reliability. Treating them all as an instruction to send creates over-mailing.

High-intent signals are explicit actions with a clear purpose: requesting a quote, booking a meeting, configuring a product, reaching checkout, renewing a subscription or downloading a detailed buying guide. These may justify a timely, narrowly relevant message.

Medium-intent signals include repeated visits to a defined solution area, comparison-page use or attendance at related webinars. They may be better used to choose the next newsletter module or to enter a short nurture path, rather than trigger an immediate sales message.

Low-intent signals include a single page view, a short visit from a search result or generic blog readership. These are generally better for aggregate content planning than individual follow-up.

Recency changes the meaning. A person viewing an event page yesterday is different from someone who viewed it six months ago. Write expiry into the rule: perhaps seven days for a checkout reminder, 30 days for a product-research audience, and a much longer period for a declared interest selected in a preference centre. The correct window comes from the buying cycle and the purpose of the message, not a universal template.

Combine website behaviour with customer context

Behaviour is strongest when it is interpreted with existing, relevant first-party information. A pricing-page visit means something different for a prospect, a trial user, a paying customer and a recently cancelled customer.

Common combinations include:

• Known subscriber + repeated product research + no conversion: send a useful comparison, implementation guide or invitation to ask questions.
• Trial user + viewed feature documentation + feature not activated: provide practical onboarding help, rather than a generic promotion.
• Customer + viewed upgrade information + account has capacity: offer relevant upgrade education, with account-owner rules where appropriate.
• Subscriber + event registration completed: exclude from acquisition emails and send logistical information.
• Customer + support journey active: suppress unrelated campaigns temporarily if the situation calls for care.

This is also where errors become most costly. Product availability, account ownership, geography, plan status and completed transactions need dependable data flows. Segments should have a named owner, a documented definition and a way to inspect a sample of included and excluded contacts before launch.

Use frequency controls to protect the relationship

Website events can happen rapidly. A person may view five pages, start two forms and return from three devices in an afternoon. That does not merit five emails.

Set global contact-pressure controls before individual trigger rules. A practical programme may give transactional and essential service messages priority, then ensure that behavioural emails respect a promotional send cap and a quiet period after another marketing email. Deduplicate events, limit one trigger per journey stage, and give the most useful current journey precedence over weaker ones.

For example, a person who has just registered for a webinar should receive confirmation and joining details. If they also qualify for a product-browsing nurture email that day, defer or suppress it. Relevance includes timing and volume.

Behavioural segmentation can support deliverability when it reduces irrelevant sending, but it is not a shortcut around poor list quality or weak consent. Watch unsubscribe, complaint, bounce and conversion patterns by segment and message type. If a narrowly targeted group produces disproportionate negative feedback, investigate the underlying assumption, the disclosure, the timing and the frequency rather than simply widening the audience.

Do not build campaigns around email opens

Email opens have always been an imperfect proxy for attention. They are less dependable as a behavioural trigger because privacy features can load tracking content without a reader actively opening the message, while other readers block remote content. Apple says Mail Privacy Protection hides recipients’ IP addresses and prevents senders from seeing whether an email was opened. Apple’s explanation of Mail Privacy Protection is a clear description of the limitation.

Use opens cautiously for high-level reporting, if at all. For individual segmentation, stronger signals include meaningful clicks, authenticated account activity, form completion, purchases, preference choices and confirmed website actions. A non-open is especially poor evidence that someone has lost interest.

Test the segment, not just the email

A strong creative test cannot rescue a faulty audience definition. Before automating a behavioural programme, validate the underlying segment:

• Review a sample of contacts and their event histories.
• Confirm that consent, preferences and suppressions are applied at send time, not only when the segment is first built.
• Check whether completed outcomes remove people quickly enough.
• Test expiry, repeat-entry and cross-device behaviour.
• Compare against an appropriate control group where volumes allow.
• Measure downstream outcomes and negative feedback, not only click rate.

Keep the first version narrow. A segment such as “all visitors to pricing” may conceal several distinct needs. A narrower starting group, such as opted-in trial users who visited a specific integration page twice in 14 days and have not activated that integration, is easier to explain, test and improve.

Where deliverability oversight fits

For established senders, a change in segmentation can alter volume patterns, message frequency and the mix of recipients at each mailbox provider. That makes deliverability monitoring useful during larger roll-outs, particularly where a new triggered programme will materially increase sending to recent subscribers or less-engaged contacts.

Email Foundry’s work is relevant where a permission-based sender needs to assess audience quality, suppression logic, contact pressure and provider-level sending patterns alongside its wider deliverability controls. The objective is not to maximise the number of triggers. It is to make sure that relevant, expected messages reach people without turning website activity into excessive or surprising email.

The enduring principle is simple: collect a small number of well-defined behaviours, explain their use, link them only to eligible contacts, give them a sensible lifespan and stop messaging once the person’s next action makes the prompt unnecessary.

Frequently asked questions

Can you email someone because they visited your website?

Not merely because they visited. An anonymous site visit is not permission to add someone to marketing. Where you have reliably linked a website event to a known contact, you still need to check that the person is eligible for the proposed email and that your tracking and profiling are transparent and lawful.

How long should website behaviour stay in an email segment?

Use a window that reflects the action. A checkout or consultation-form start may lose value within days, while a declared content preference may remain useful for much longer. Document the expiry period and remove or downgrade stale signals rather than treating old behaviour as current intent.

Is form abandonment a good email trigger?

It can be, but only for an appropriate form and a known, eligible contact. Distinguish a form start from a completed request, suppress the reminder as soon as the form is submitted, and avoid repeated prompts. Support, complaint, sensitive or high-friction forms need particular care.

Should open rate be used for behavioural segmentation?

It is a weak signal for individual targeting. Privacy tools and blocked images make opens unreliable. Prefer meaningful site actions, clicks, form outcomes, product use and explicit preferences, while applying frequency controls to every trigger.

Sources and further reading

• Cookies and similar technologies — Information Commissioner's Office
• Collect information and generate leads — Information Commissioner's Office
• Identify direct marketing — Information Commissioner's Office
• Data protection impact assessments — Information Commissioner's Office
• Data minimisation — Information Commissioner's Office
• Use Mail Privacy Protection on iPhone — Apple Support