Website Security Monitoring vs One-Time Scanning
Use a one-time scan if the app will not change after the scan. Use monitoring if the app holds user data and changes each week. Most live apps need one scan first and monitoring after it.
Both options use the same kind of test. The difference is time. A one-time scan shows the app on one day. Monitoring repeats the test and shows what changed. This page helps you to choose. For the features of monitoring, see continuous web application security monitoring.
What does each option do?
| One-time scan | Monitoring | |
|---|---|---|
| Runs | Once, when you start it | Again and again, with a comparison |
| Shows | The state of the app today | The change since the last run |
| Good at | A baseline, a launch check, a fix check | Finding a new problem after a change |
| Weak at | Anything that changes after the run | Business rules and logic problems |
| Effort | Read one report | Read each change, and decide |
A one-time scan is not a bad option. It is the correct option for some apps. Monitoring is also not a full answer. It needs a person to read the changes and act on them.
What does a one-time scan miss?
A scan has a blind time. This is the time from the end of the scan to today. Each event in that time is not tested.
These events are common.
- A deploy adds a table without an access rule.
- A host change removes a security header.
- A person turns off a policy in a dashboard to find a bug, and forgets to turn it on.
- A developer puts a secret in client code.
- A preview deploy or a staging host goes online with real data.
- A vulnerability in a package that you already ship becomes public. The CISA Known Exploited Vulnerabilities catalog shows how often new items are added.
An attacker does not need skill for these cases. The scan only needs to be old. If the blind time is long, more of the life of your app is not tested.
An app that an AI editor changes has a short life between changes. Each prompt can change code, database rules, or both. For these apps, the blind time grows fast.
How long can a problem stay open?
The worst case is the full time between two runs. Use this table to see the effect.
| Scan frequency | A problem that starts just after a run can stay open for |
|---|---|
| Once, at launch | Until the next time someone scans |
| Every three months | Up to about three months |
| Every month | Up to about one month |
| Every week | Up to about one week |
| After each deploy | Until the next deploy and its result |
VibeEval monitoring re-checks a target every week. A scan after each deploy needs a manual re-run. See the monitoring guide for the steps. Monitoring is an early-access feature, so use manual re-runs for releases that matter.
The table shows the maximum time. Many problems stay open for less time. But you do not know in advance which problem will stay open.
How do the costs compare?
Do not compare only the price of a run with the price of a plan. Compare the whole cost.
| Cost item | One-time scan | Monitoring |
|---|---|---|
| Money | One run | A recurring plan |
| Time to start | Low | Low, plus one baseline pass |
| Time each month | None | Time to read changes |
| Cost of a late-found problem | Can be high | Lower, because you find it sooner |
| Cost of noise | None | Real, if you do not control alerts |
Current plan prices are in the pricing section. The plan limits are in plans and limits. The Pro plan includes 100 scans each month and 50 targets. That is enough for a weekly re-check of a small set of apps.
Think about the cost of a leak. It includes the time to find the cause, to tell users, and to fix the problem. It also includes lost trust. The IBM Cost of a Data Breach report gives average figures for breaches. Use it to judge the scale. Do not use one average for your own case.
A simple test helps. Ask what happens if user data is public for one month. If the answer is “nothing much”, a one-time scan may be enough. If the answer is “we lose customers”, use monitoring.
When is one scan enough?
Choose a one-time scan if all of these are true.
- The site is static, or it has no accounts and no stored user data.
- You will not change the site. Or you will scan again by hand after each change.
- The app is a prototype, a demo, or a short event site with an end date.
- You test before a single event, such as a launch, and you will test again before the next one.
In these cases, monitoring adds cost and gives little more. Put your effort into fixing the findings from the one scan.
When do you need monitoring?
Choose monitoring if one or more of these are true.
- User data is behind the app, for example emails, payment data, or documents.
- You deploy each week or more often.
- An AI editor or an agent can change production code.
- Many people can edit database rules, DNS, or hosting settings.
- You run more than one environment, and previews can stay online.
- A customer or a partner asks how you find regressions.
If you manage many sites for clients, read multi-site security monitoring. It covers client reports and triage.
Decision table
| Your situation | Choose |
|---|---|
| A static site with no accounts | One-time scan |
| A prototype that you will delete | One-time scan |
| Before launch, no users yet | One-time scan and the pre-launch checklist |
| A live app with accounts and weekly deploys | Monitoring |
| A Supabase or Firebase backend that an AI editor changes | Monitoring |
| Many editors or many environments | Monitoring |
| Pricing, roles, or approval flows to verify | Monitoring and a periodic pentest |
| A release gate on each deploy | A CI test, see smoke testing |
How do you make the decision?
Follow these steps.
- Run one scan on your production URL. The free self-audit or the vibe code scanner can give you a baseline.
- Fix the critical and high findings. Run Retest on each one.
- Count how many times the app changed in the last month. Include code, database rules, and hosting settings.
- If the app changed more than once a week, or it holds user data, turn on monitoring.
- If the app did not change and holds no user data, keep the one scan. Write the date in your calendar and scan again before the next change.
- Add a pentest when the app handles money or sensitive records.
You can change your decision later. Many teams start with one scan and add monitoring after the first real user signs up.
What does this look like in a real week?
Use an example. A small team runs a SaaS app on Supabase. They scan before launch and find no critical findings. They stop there.
- On Monday, an AI editor adds a
notestable for a new feature. The prompt does not mention access rules. - On Tuesday, the team deploys. The table has no row-level security policy.
- On Wednesday, any visitor with the public API key can read every note.
- The next scan happens three months later, if it happens at all.
With a weekly re-check, the same table appears as a new finding at the next run. The team reads the evidence, adds the policy, and uses Retest to confirm the fix. The problem stays open for days, not months. With a re-run after the deploy, it stays open for hours.
The example is simple on purpose. Most real problems are as simple as this one. They come from a normal change that no person examined for security.
What should you ask before you pay for monitoring?
Ask each question. A good answer to all of them means that the tool fits.
- What does it test? Make sure that it tests the deployed app from outside, with a login if your app needs one.
- How often does it run? Make sure that the schedule matches how often your app changes. State if the feature is early access.
- How does it show a change? Make sure that you get a comparison, not only a new list of all findings.
- How does it send an alert? Make sure that it supports email and a webhook for your tools.
- Who reads the result? Name a person. A tool without an owner gives no value.
- How do you verify a fix? Make sure that you can run a single check again, not only a full scan.
If you cannot name an owner for question 5, do not buy monitoring yet. Use a one-time scan, and add monitoring when a person has time to act on the alerts.
Which option is better for an agency or a small business?
An agency or a small business often has many sites and few people. The risk is different from the risk of one SaaS app. Many sites are static, which favors a one-time scan. Some sites hold customer data, such as a shop or a booking form, which favors monitoring.
Sort the sites into two groups. Put the static sites in the first group and scan each one at launch and after each change. Put the sites with user data in the second group and monitor them. This keeps cost and effort low. For client reports and triage across many sites, see multi-site security monitoring.
What does monitoring not fix?
Monitoring has limits. Plan for them.
- It does not stop an attack. It finds a weakness.
- It does not test business logic as a person does. See vulnerability scanning vs AI pentest for the difference between scanners and pentests.
- It tests only the targets you register. A forgotten host stays outside the scan. Use the guide on discovering forgotten web apps to find such hosts.
- It does not fix a problem. A person or a coding agent must make the change.
A one-time scan has the same limits, and also has the blind time described above. The choice between the two is a choice about that blind time.
NIST gives the wider view in SP 800-137. It says that continuous monitoring supports risk decisions with current data. For one web app, you can apply the same idea in a small way.
What do we not say?
- We sell monitoring, so we have a reason to say that you need it. The table sends static sites and short-lived prototypes to one scan, because that is the correct answer for them.
- Monitoring in VibeEval is an early-access feature that re-checks targets each week. It does not give a result in real time.
- No scan, one-time or repeated, proves that an app is secure. A scan shows the problems it can find.
Related
Common questions
Is one security scan enough?
How long can a problem stay hidden between scans?
Does monitoring cost more than a one-time scan?
Can I use both?
What should a solo founder do?
Does monitoring replace a pentest?
Start with one scan
Run a baseline scan on your URL. If the app keeps changing, turn on monitoring for weekly re-checks. 14-day trial, no card.
14-day free trial · Cancel anytime