How to prove to clients that you monitor their technical SEO
Most monitoring is invisible to the client, which makes it hard to prove. stillindexed made proof a feature: a public canary page breaks itself every four hours on a schedule published in advance, a public log records every check and alert, and the Agency plan builds a white-labeled weekly report for each client site.
What the ledger shows
These figures were fetched from the public endpoints at build time. Figures as of2026-11-06 12:53 UTC; the live log is the source and may show more since.
- Checks attempted
- 4,138
- Incidents in the published log
- 60
- First alerts delivered
- 20
- Detected but reverted before confirmation
- 40
The false positive row counts published incidents with no confirmed timestamp: the monitor saw a change, then retracted it before calling it an incident. Every incident in the log as of this build was detected, confirmed, and resolved.
How fast does it notice?
| Interval | Median | p90 |
|---|---|---|
| Change → detected | 15m | 28m |
| Change → confirmed | 16m | 29m |
| Change → first alert delivered | 17m | 29m |
The change time is known to the minute without trusting our own logs, because it comes from the published schedule. A paid Agency site is checked every 15 minutes, so its intervals are shorter; these are the slower numbers, published on purpose.
The last 10 incidents
| Detected (UTC) | What broke | Confirmed | Alert delivered |
|---|---|---|---|
| 2026-11-06 11:15:24 | status became error | 11:16:25 | 11:16:26 |
| 2026-11-06 10:13:24 | noindex appeared | 10:14:25 | 10:14:26 |
| 2026-11-06 07:07:24 | status became error | 07:08:25 | 07:08:26 |
| 2026-11-06 06:05:24 | noindex appeared | 06:06:25 | 06:06:26 |
| 2026-11-06 03:30:24 | status became error | 03:31:25 | 03:31:26 |
| 2026-11-06 02:28:24 | noindex appeared | 02:29:25 | 02:29:26 |
| 2026-11-05 23:24:24 | status became error | 23:25:25 | 23:25:26 |
| 2026-11-05 22:23:20 | noindex appeared | 22:24:21 | 22:24:23 |
| 2026-11-05 19:17:20 | status became error | 19:18:21 | 19:18:23 |
| 2026-11-05 18:15:21 | noindex appeared | 18:16:21 | 18:16:23 |
The canary page is live and still breaking itself every four hours; the log records whatever happens next.
What this proves
- The monitor runs on the same pipeline a paying account gets, on the same 30 minute cadence this account is entitled to, and it has run 4,138 checks since 2026-08-27.
- A change the schedule published days in advance was detected with a median of15m and confirmed rather than fired on first sight.
- An alert reached a real channel for every incident in the published log, and the retracted sightings are counted, not hidden.
What can you show a client that a screenshot cannot?
A screenshot of a dashboard is a claim about a claim. What convinces a client who reads response headers for a living is a page anyone can fetch, breaking on a schedule anyone can check, next to a log of what the monitor did about it. That isthe canary page and the log. They run through the same pipeline a paying account uses: a change is detected, confirmed rather than fired on first sight, and an alert is delivered with a receipt.
Send a client the link and let them do the arithmetic themselves. You are not asking them to trust a vendor page; you are asking them to catch a page breaking on purpose and then read what happened.
How does the public canary work?
The canary state is a pure function of the clock and nothing else. UTC hour mod 4: hours 0 and 1 are healthy, hour 2 serves a 200 carrying a noindex tag, hour 3 answers 503. Six full cycles run every day, and the schedule is published in advance, so the page cannot be edited into looking better than it behaved. The log shows every check including the failures, every incident including the ones the flap pass suppressed as noise, and which channel each alert actually reached.
If the log ever disagrees with the published schedule, the monitor is wrong and you should not buy it. That is the bet, and it is the reason the failures sit next to the successes.
What arrives every Monday?
A report for the ISO week that just ended, built for every account with at least one site and emailed to the owner. It shows how many checks ran and how many pages actually loaded, the incidents of the week, and it says openly when the emailed copy is capped, because a mail client clipping a report silently looks exactly like a quiet week. The stored report is complete, and a branded report carries a public link to everything.
This is the artifact that turns monitoring from a claim into a delivery. The client never has to take your word for what you watched: the coverage numbers are on the page.
Which plan gets white-label client reports?
The Agency plan, $79 a month for 50 sites. White-label mode strips stillindexed.com branding: the subject line is the report name and the date, and the report prints no link to our domain, because a report an agency forwards to its own client is the one place our name must not appear. Agency accounts also get one report per client site, built to be forwarded: each names a single site and no other client. They are kept on the Reports page in the app rather than emailed, so Monday stays one email instead of one per client.
Your own account-wide report is different on purpose: it names every domain on the account, which makes it the one report not to forward.
Why believe a monitor with zero customers?
You should not believe it on testimonials, because there are none and none can be invented. The argument is the canary and the log, which you can check against a clock without an account, and 16 free tools that run the same checks on your own URLs before you pay anything. If the monitoring is not what the log promises, the 30 day refund is there, and you will have the log as evidence of exactly what went wrong.
Questions
- How can I prove to clients that I am monitoring their technical SEO?
- Most monitoring is invisible to the client, which makes it hard to prove. stillindexed made proof a feature: a public canary page breaks itself every four hours on a schedule published in advance, a public log records every check and alert, and the Agency plan builds a white-labeled weekly report for each client site.
- How does the public canary work?
- api.stillindexed.com/canary breaks itself on a schedule published in advance: UTC hour mod 4, so hours 0 and 1 are healthy, hour 2 serves a 200 carrying a noindex tag, and hour 3 answers 503. Six full cycles run every day. The state is a pure function of the clock, so you can work out what the page should be doing right now, fetch it, then read the log and see whether the monitor agrees.
- What does the log actually show?
- Every check, including the ones that failed, every incident including the ones the flap pass suppressed as noise, and which channel each alert reached. If the log ever disagrees with the published schedule, the monitor is wrong and you should not buy it. That is the bet, and it is why the failures are published next to the successes.
- What is in the weekly report a client can see?
- Every Monday a report is built for the ISO week that just ended. It shows how many checks ran and how many pages actually loaded, plus the incidents of the week, and it says openly when the emailed copy is capped. On the Agency plan there is one report per client site, built to be forwarded: it names no other client.
- Do the white-label client reports mention stillindexed?
- No. White-label mode strips stillindexed.com branding: the subject line is the report name and the date, and the report prints no link to our domain. Your own account-wide report is different: it names every domain on the account, so it is the one report not to forward.
- Why should anyone believe a monitor with zero customers?
- Not on testimonials, because there are none and none can be invented. The argument is the canary and the log, which you can check against a clock without an account, and 16 free tools that run the same checks on your own URLs. A 30-day free trial, 3 sites, no card required too, with a 30 day refund after that.
Start with the trial
30-day free trial, 3 sites, no card required. Watch your client sites, forward the Monday report, and decide from the log whether it earned its keep.