Website Changes

A Website Changes trigger monitors a defined set of company domains and runs your instructions when Swan’s data provider reports a website change that matches your natural-language filter. Use it for focused monitoring such as competitor pricing and packaging, product launches, security or compliance pages, partnerships, careers pages, or changes at open-deal accounts. This is different from Website Visitors, which identifies people or companies browsing your own site, and Business Events, which monitors structured company events such as funding or leadership changes. Create one from Triggers → New trigger → Website Changes. Website Changes is an organization-wide trigger rather than a personal trigger.

​

Choose which companies to monitor

A trigger needs at least one company source. You can combine:

• Explicit company domains, including domains imported from a CSV.
• HubSpot company lists. Swan resolves the lists’ current members each time it checks.
• Filters over companies already in Swan, including revenue, employee count, owner, funnel stage, target market, and tags.

Swan resolves and checks at most 1,000 domains per trigger poll. If your combined sources contain more, narrow the cohort; domains beyond the cap are not checked during that poll. When a provider-matched domain does not already have a Swan company record, the first check creates one with the trigger as its source. A company on your excluded-companies list may still get a check record, but its change is not classified and does not run the playbook.

​

Define the change filter

The change filter says which observable website changes should qualify. It runs before your trigger instructions, so keep actions such as research, Slack notifications, CRM updates, and outreach in the instructions instead. A useful filter names both the signal and nearby noise to ignore. For example:

Pricing, packaging, plan limits, or free-trial changes. Ignore cosmetic redesigns, copy-only edits, and customer-logo updates.

Swan compares the filter with the provider’s changed URL, change description, and implication. It does not inspect raw page HTML or receive a complete text diff at classification time.

​

Cadence and change history

Website Changes can run daily, weekly, or monthly, at a time you choose. It cannot check more often than once per day, and the schedule uses UTC. On the first poll, Swan may evaluate the latest change already known to the provider; the change is not guaranteed to have happened after you enabled the trigger. After Swan successfully classifies a provider-reported change, it remembers that provider business and change date for one year for that trigger configuration. Later polls with the same date are marked unchanged. Editing the change filter does not replay an already-classified dated change.

​

What happens during a check

For each company the provider can match, Swan records a website-change check and gives it one of these outcomes:

• Missing data — the provider had no usable website-change data for the company. No classification or agent run occurs.
• Unchanged — Swan already evaluated the provider’s current change date. No classification or agent run occurs.
• Company excluded — the company is excluded in Swan. No classification or agent run occurs.
• Filtered out — the provider reported a new change, but the classifier decided it did not match your change filter. The downstream playbook does not run.
• Matched — the classifier decided the new change matched your filter, so Swan starts the configured playbook with the domain, changed URL, date, description, implication, and classification reason.

These outcomes let you distinguish a quiet website, missing provider coverage, an overly narrow filter, and a downstream instruction problem. See Monitoring runs for where to inspect execution history. If the provider cannot match an enrolled domain to a business at all, Swan cannot request website-change data for it, so that domain has no check record for the poll. This is different from Missing data, where the business matched but had no usable change data. Matched downstream agent runs start at up to 10 per minute per organization. Additional matches wait in the queue rather than being discarded.

​

Credits

Classification costs 1 credit for each new provider-reported change Swan evaluates, whether the result is Matched or Filtered out. Missing data, Unchanged, and Company excluded checks cost 0 credits. When a change matches, the downstream playbook’s actions cost their normal credits in addition to the classification credit. Classification happens before the downstream agent run, so its credit can still be charged when the later run cannot start or the trigger becomes inactive before the playbook begins. See the canonical Credit costs.

​

Testing before rollout

Trigger simulation tests a hypothetical already-matched event against your downstream instructions, skills, permissions, and tool choices. It does not call the website-change provider or run the production classifier, so a successful simulation does not prove that a real page change will be detected or match your filter. Before expanding a watchlist:

1. Review the domains, lists, filters, cadence, change filter, and downstream instructions.
2. Simulate realistic matched scenarios, including customer, prospect, active-deal, and unavailable-destination branches where relevant.
3. Enable the trigger for one representative domain and inspect real check outcomes before adding the full cohort.

There is currently no manual replay or classifier-preview path for a past website change.

​

Common questions

Can this monitor every page on a website in real time? No. Swan checks on a daily-or-slower schedule and evaluates the latest change summary supplied by its data provider. It is not a real-time crawler or a complete page-by-page diff service. Why did changing my filter not re-evaluate the last change? Deduplication is based on the provider business and change date for the trigger configuration. Once that date has been classified, editing the filter does not make it new again. Why do I see a check even though no agent ran? Swan records checks for provider-matched companies so you can see whether data was missing, unchanged, excluded, or filtered out. Only Matched starts the downstream playbook. Why does an enrolled domain have no check in history? The provider may not have matched that domain to a business. Missing data appears only after a business match succeeds; a domain that cannot be matched has no per-company check to record. Does a filtered-out change cost credits? Yes. The credit pays for evaluating a new provider-reported change against your filter. Quiet, missing-data, unchanged, and excluded checks are free.