SaaS Pricing Monitor — Plans, Change Alerts & Battlecard

Monitor competitor SaaS pricing pages and get structured plans, typed change alerts, and a decision-ready battlecard — not a wall of page text.

Point it at pricing URLs, or just give it a keyword ("project management software") and it auto-discovers the top competitor pricing pages itself.

Why this monitor

Most pricing trackers regex the page text and dump it. You get junk plan names, prices with no plan attached, and "changes" that are just copy edits. This actor extracts the actual plan cards:

{
  "name": "Pro",
  "amount": 8.75,
  "currency": "USD",
  "period": "month",
  "per_seat": true,
  "is_free": false,
  "is_custom_price": false,
  "features": ["Unlimited message history", "Group video calls", "..."],
  "cta": "Get Started"
}

…and diffs them statefully between runs, so you get typed events, each one answerable with an action:

Event	Fires when
price_changed	a plan's price moved — with old, new, % and direction
plan_added / plan_removed	the plan lineup changed
trial_changed	free trial appeared, disappeared, or changed length
annual_discount_changed	annual-billing discount appeared or changed
billing_period_changed	a plan switched monthly ↔ annual billing
features_changed	a plan's feature list materially changed
page_changed_unparsed	page changed but nothing structured moved (honest signal, never silent)

The first run on a URL bootstraps a baseline (no change events, no surprises). Every scheduled run after that emits only what changed.

Keyword auto-discovery (no URL paste)

Set keyword to your category and the actor finds the top competitor pricing pages via search (Google SERP when your account has the GOOGLE_SERP proxy group, otherwise Bing/DuckDuckGo), filters out review sites and "best X software" listicles, verifies each candidate is a live pricing page, and monitors them.

You can mix modes: pin your must-watch competitors in pricingPageUrls and let discovery fill the remaining slots up to maxCompetitors.

Battlecard

Each run can emit one competitive battlecard across all monitored pages:

• cheapest & priciest paid entry points, median entry price
• free-tier and trial matrix per competitor
• full plan matrix (name, price, period, per-seat)
• everything that changed this run, summarized
• positioning notes ("X has no trial and no free tier — friction opportunity")

Pipe it to Slack or your CRM via a webhook on the run, or schedule the actor weekly and read the battlecard as your Monday pricing brief.

Input

Field	Type	Notes
keyword	string	SaaS category for auto-discovery. Optional if URLs given.
pricingPageUrls	array	Public pricing page URLs. Optional if keyword given.
maxCompetitors	integer	Cap on pages per run (default 10, max 25).
emitBattlecard	boolean	One battlecard per run (default true).
eventTypes	array	Filter which change events to emit. Empty = all.
kvStoreName	string	Named KV store for state. Keep it stable across runs.
forceBrowser	boolean	Force stealth-browser rendering (rarely needed — escalation is automatic).
proxyConfiguration	object	Apify Proxy; defaults to residential.

Output records

All records land in the default dataset with a record_type discriminator:

• snapshot — one per monitored page per run: plans, trial/discount signals, parse_confidence, extraction_method
• change_event — typed events as in the table above
• battlecard — one per run

Dataset views (Snapshots / Change events / Battlecard) are pre-configured in the Apify console.

How it parses (and why it's reliable)

1. Fast static fetch with TLS impersonation; JS-shell pages escalate automatically to a stealth Camoufox browser render.
2. Plan cards are found structurally — repeated sibling card patterns in the DOM, not class-name guessing, not page-text regex. JSON-LD offers enrich the result when present.
3. Junk grids (FAQ accordions, feature-comparison tables) are rejected by per-grid quality scoring.
4. Every snapshot carries parse_confidence and extraction_method, and a degraded parse never fabricates plan-removal events — you get an honest page_changed_unparsed instead.

Pricing events

Event	Charged
pricing-snapshot	per pricing page successfully extracted
pricing-change-event	per typed change event emitted
battlecard-report	per battlecard generated

Failed pages are never charged. Baselines are charged as snapshots only — a first run never bills change events.

Scheduling

Typical setup: a weekly (or daily) schedule with your competitor set, a stable kvStoreName, and a webhook or integration reading the dataset. The actor is stateless between runs except for the named KV store, so you can fork configurations per product line by changing kvStoreName.

Legal

Only public pricing pages — no login, no paywall, no personal data. You own your runs and the extracted data. Monitoring public competitor pricing is standard competitive-intelligence practice; this actor reads nothing a regular visitor's browser wouldn't.