The question this example tests
You want to know when a plan's price or terms change. A rotating announcement at the top of the page keeps changing too. Should every announcement reach the person watching the price?
This controlled example separates the two signals. Three fictional Fieldnotes pages contain the same plan region. The second changes only an announcement. The third also raises the monthly amount from $29 to $39. The plan still includes five seats, is priced in USD and excludes tax.
The captures are actual browser output from authored fixtures. They do not measure a production website's alert volume or claim a percentage reduction in false positives.
What changed in the captures
Scroll horizontally to see all columns.
| Comparison | Whole page | Selected plan region | Review decision |
|---|---|---|---|
| Baseline to new announcement | Screenshot and text change | Screenshot bytes and text remain equal | Quiet for this price-monitoring purpose |
| New announcement to new price | Screenshot and text change | Monthly amount changes from $29 to $39 | Review the plan change |
The same observations hold at the two tested widths: 1440 and 390 CSS pixels. The region is selected with #team-plan. It contains the plan name, amount, period, seat quantity, tax note and action, so the price is not stripped of its meaning.
The plan's monthly amount increases by $10. Dividing $10 by the original $29 gives approximately 34.5%. That calculation assumes the same billing terms and included quantity; it does not include tax or an annual commitment.
Choose a region that keeps the decision intact
A narrow crop can remove useful context. Selecting only the number would miss a change from monthly to annual billing. Selecting only the price card may miss an important condition introduced immediately below it. Start with the question you need to answer, then include every nearby term required to interpret the answer.
For a pricing monitor, check that the region contains the correct plan, currency, billing period, commitment, quantity and availability. Read the excluded area before calling it irrelevant. A promotional banner that changes the offer's terms belongs in the evidence, even when its wording rotates frequently.
Write down the exclusion reason: “Announcements about documentation and workshops do not affect the Team offer.” That is narrower and easier to review than “Ignore the top of the page.” Revisit the choice after a redesign or an offer change.
Handle a missing region as missing evidence
If #team-plan disappears, the comparison has lost its intended subject. It has not established that the price stayed the same. Record the match count, preserve the capture outcome and investigate whether the plan moved, was renamed, was removed or failed to load.
The companion selector identity example shows another failure: a selector can still match one element while silently switching to a different plan. Both identity and match count matter.
Reproduce the example
Download the monitoring lab kit, extract it, then use Node 20 or later and an installed Chrome browser:
npm ci --ignore-scripts
node scripts/capture-monitoring-lab.mjs
Compare banner-before, banner-rotated and banner-price-changed in the generated public/evidence/monitoring-lab/results.json. The region records identify the selector and screenshot hashes. full_text and region_text preserve a readable comparison.
You can also inspect the published baseline fixture, announcement-only fixture, price-change fixture and recorded results.
What this result does not establish
The announcement is represented by separate deterministic states. The harness does not run a timed carousel, sample rotation frequency or estimate a real alert rate. Its screenshot equality is exact byte equality within one recorded browser run, not a perceptual-difference threshold that applies to arbitrary pages.
Region monitoring covers the chosen region. It does not establish the accessibility, accuracy or stability of the rest of the page. A wider periodic review can be useful when excluded content has business importance.
Put the scope in writing
Use the monitoring brief to record the decision, region, exclusions, missing-target behavior and review owner before turning on alerts. For the broader workflow, read reducing false-positive website alerts.