Start with the job
Three jobs decide the answers: knowing before the client does, showing the client the work, and catching the quiet breakages that “up” hides (an expiring certificate, a changed DNS record, a page that became noindex). If your real job is on-call response for your own application, an incident tool fits better and this list is the wrong one.
The twelve questions
1. How does it decide a site is down?
This sets how many false alarms you will answer. A single failed check from one place is often the path between the monitor and the site. Tools differ: Oh Dear re-checks from a second, randomly chosen location and needs both to agree, UptimeRobot re-checks up to three times from the same region and, with several regions, marks a monitor Down when any single region fails three consecutive checks, and some tools say only that they avoid false positives. The rules, with sources, are in Two-region confirmation and false alarms. Ask for the rule in writing and for what the alert shows.
2. Can it group sites by client?
A portfolio is organised by client, not by monitor. Look for clients or groups, a report per client, a status page per client and alerts that can go to a different person for each client. UptimeRobot’s agency page describes monitors organised by client with tags; Oh Dear’s pricing page describes grouping each client’s sites with their own monthly reports.
3. What does the client receive?
Most retainers need three things: an alert to you, a monthly report the client can read, and sometimes a public status page. Check who sends each, what it contains and whether gaps in measurement are shown honestly or counted as up. See How to write a monthly website care report and Public status pages for agencies.
4. How white is the label?
“White label” covers six different things, and vendors price them separately. Decide which ones your clients notice before you compare plans: White-label website monitoring for agencies.
5. What does it watch besides “is it up”?
List the last five incidents across your portfolio and write down which check would have caught each. Typical gaps are a certificate that expired, a domain that lapsed, a DNS record that changed, a redirect that broke, a robots.txt that blocked everything or a noindex that shipped with a deploy. Then check which candidate has that check on the plan you would buy, not only on a higher one.
6. How do alerts reach the right person?
Check the channels (e-mail, Slack, webhook, SMS or phone), whether routing can differ per client, whether a new alert address must be confirmed, and what happens out of hours. An alert that nobody owns at 3 a.m. is a false sense of safety.
7. How fast, and from where?
A shorter interval detects sooner and, with a confirmation rule, costs little in false alarms. The check interval sets the average delay (half the interval) and the worst case (the whole interval). Google’s SRE book makes the opposite point: for a service allowed no more than 9 hours of downtime a year (99.9 percent), probing for a 200 status more than once or twice a minute is “probably unnecessarily frequent”. Pick the interval from what a minute of downtime costs your client, then ask how many locations the check runs from and whether you can choose them. More background: How often should you check a website.
8. What does the price count?
This is where comparisons go wrong. Plans count different things, so compare the unit before the price.
| Tool | What the plan counts | Source |
|---|---|---|
| UptimeRobot | Monitors | UptimeRobot pricing |
| Better Stack | Monitors and heartbeats, added in blocks | Better Stack pricing |
| StatusCake | Uptime monitors, with separate counts for page speed, domain, SSL and server monitors | StatusCake pricing |
| HetrixTools | Uptime monitors | HetrixTools uptime pricing |
| Oh Dear | Sites and servers | Oh Dear pricing |
| Site24x7 | Websites, with transaction monitors and status pages counted apart | Site24x7 pricing |
| Pingdom | Uptime checks and advanced checks, with SMS counted in each tier | Pingdom pricing |
| Little Warden | URLs patrolled | Little Warden pricing |
Read on 2026-10-28. To compare, convert everything to the cost of one client site: the plan price divided by the number of client sites it really covers. If a tool counts monitors, count the monitors one client site needs. A site with a page check, a certificate check and a DNS check is three monitors, so a plan of 100 monitors covers 33 such sites, not 100. If a tool counts URLs, count the pages you want watched on each site. Prices change and some pages show different numbers by region, so take the numbers from each vendor’s page on the day you decide.
9. Who can sign in?
Check seats, roles, two-factor sign-in and single sign-on, and whether a client can have read-only access or receive alerts without a paid seat. Oh Dear’s pricing page describes read-only access to just a client’s sites; UptimeRobot’s agency page describes notify-only seats for clients. If your clients only read a PDF, none of this matters.
10. What happens to the history?
Ask how many months are kept on the plan, whether you can export it, and what is deleted when you cancel. A monthly report is only as good as the stored checks behind it. Export a sample before you commit.
11. How do you try it, and how do you leave?
Compare the free period, the card requirement and the money-back promise, and then read how to cancel and export.
| Tool | Free period or plan | Money-back | Source |
|---|---|---|---|
| Oh Dear | 10 days with every feature, no card | 30 days for new customers | Oh Dear pricing |
| UptimeRobot | Free plan with 50 monitors and 5-minute checks | Not stated in the plain-text pricing | UptimeRobot pricing |
| Better Stack | Free plan for personal projects; no free period stated | 60 days | Better Stack pricing |
| StatusCake | 7 days on any plan, no card; a free plan | Not stated | StatusCake pricing |
| Little Warden | 40 days free, no card | 30 days | Little Warden pricing |
| Site24x7 | 30 days, no card; a free plan for up to 50 resources | Not stated | Site24x7 pricing |
| HetrixTools | A free plan (log in at least every 90 days); no free period | Not stated | HetrixTools uptime pricing |
Read on 2026-10-28; terms can change.
12. Who runs it, and how do you know?
Look for the operator’s company details, a data processing agreement, the hosting region, the vendor’s own status page and a changelog. A newer tool has less history to inspect, so ask for what exists: the legal pages, the security notes, and how long the vendor has published its own incidents.
A 30-minute trial test
Run the same test on every candidate and keep the notes in one table.
- Add three real sites of different stacks, in bulk if the tool allows it.
- Break one of them on purpose on a staging copy: serve a wrong DNS answer, a noindex or a robots.txt that blocks everything. Time how long the alert takes and read what it shows.
- Read the alert. Does it say what each location saw, and what changed just before?
- Generate the client report and read it as the client would. Are gaps shown as gaps?
- Open the status page signed out. Check what it shows during the broken state.
- Count the false alarms over a week on a site you know is fine.
- Export your data once and read the cancellation terms.
Score each of the twelve questions 0 (not offered), 1 (partly) or 2 (as you need), count the first five twice, and write down the notes next to the score. The total decides little by itself; the zeros tell you what you would have to live without.
How Beamfar helps
Beamfar is built around the first five questions: a site is down only when two regions agree, sites group by client from the Solo plan, each client gets a monthly report and a status page, your own name goes on both on the Agency and Agency Plus plans, and certificates, domain expiry, DNS records, redirects and SEO-risk signals (noindex, robots.txt blocking, a canonical moved to another host) sit on one timeline. It is narrower elsewhere: checks run every minute (every 30 seconds on Team and up, as a choice per check), with no SMS or phone alerts, a Free plan of one site only after the trial, and no reseller billing; a read-only client portal is on the Team, Agency and Agency Plus plans. It is new, with no paying customers yet, so question 12 has no track record to show yet. The comparison pages set it against other tools, with sources and the date read.
Questions
What should an agency look for in website monitoring?
How downtime is confirmed, whether sites group by client, what the client receives each month, how much of the label is yours, what is watched besides uptime, and what the plan counts when it sets the price. Test these in a trial before you commit.
Is a free uptime monitor enough for client sites?
It can be for a handful of brochure sites. UptimeRobot, for example, lists a Free plan with 50 monitors and 5-minute checks (read 2026-10-28). What decides whether it is enough for a portfolio is the confirmation rule, client grouping, the report you can give the client and the checks beyond uptime.
How many monitors does one client site need?
It depends on what you want watched. A page check, a certificate check and a DNS check make three monitors for one site in a tool that counts monitors, and each cron job, API or port check adds one more.
Should I pay for 30-second checks?
Only if a minute of downtime costs your client more than the difference in price. Beamfar offers 30-second checks on Team and up as a choice per check; a new check runs every minute. For a service allowed 9 hours of downtime a year, Google’s SRE book calls probing more than once or twice a minute probably unnecessarily frequent.
Sources
Read on . Standards and vendor pages change; confirm the current text before you rely on it.
- Google SRE book, Monitoring Distributed Systems (sre.google)
- Oh Dear documentation, uptime and performance monitoring (ohdear.app)
- UptimeRobot help, locations and multi-location (help.uptimerobot.com)
- Oh Dear pricing (ohdear.app)
- UptimeRobot pricing (uptimerobot.com)
- Better Stack pricing (betterstack.com)
- StatusCake pricing (www.statuscake.com)
- HetrixTools uptime pricing (hetrixtools.com)
- Pingdom pricing (www.pingdom.com)
- Little Warden pricing (www.littlewarden.com)
- Site24x7 pricing (www.site24x7.com)