SilentShield
← Back to Blog
ReleaseDashboardTransparency

SilentShield 2.7: see how every verdict came about

Marc Wagner·October 12, 2026·3 min read

SilentShield 2.7 answers a question we hear often: why did this visitor get this verdict? Every entry under User Interactions now has a page that walks through the decision step by step. The evaluation itself does not change; the page shows how it came about. The full list is on the changelog.

What is new

Entry page with the verdict “Suspicious, 0.42”, the score on a scale from 0.00 to 1.00 and, at the top right, the button “How did this verdict come about?”.
The button sits next to the verdict on every entry. The screenshot is in English; the dashboard follows your language.

Open any entry under User Interactions in your dashboard. Next to the verdict there is a new button, “How did this verdict come about?”. It opens a page with seven steps on the left and the detail of the selected step on the right. Each step carries a status, so you see at a glance where something stands out.

What the seven steps show

The page “How did this verdict come about?” with the seven steps on the left and step 3, Signals, selected: keyboard, typing pauses, mouse, pointer speed, visibility and focus, scroll behaviour and browser capabilities, each with a bar, a value and the rating weak, medium or strong.
Step 3 in the customer view: every signal with its value and strength. The screenshot is in English.
  • Entry: the key, the page and the form the visitor used.
  • Blocks: whether a honeypot was filled and whether the visitor came through Tor or a data center.
  • Signals: the seven behavior values the widget measures, each rated from 0 (bot-like) to 1 (human) and shown as weak, medium or strong.
  • Reasons: what lowered the score further, for example no typing pauses or a repeated fingerprint.
  • Limits: where the score sits between the bot limit and the human limit, and whether the limit decided.
  • Policy: whether one of your own rules moved the verdict.
  • Verdict: the result.

If a rule decided before the score counted, such as a filled honeypot or a repeated fingerprint, the page names that stage. Numbers are written the way your language writes them: 0,40 in German, 0.40 in English.

What you do not see, and why

Step 5, Limits, selected: a bar from bot to human with the marker “This entry 0.42”, the bot limit 0.20 and the human limit 0.70, and the sentence that the score is above the bot limit and below the human limit, so the verdict is suspicious.
Step 5 shows where the score sits between the limits. Weights and deduction amounts do not appear in the customer view. The screenshot is in English.

As a customer you see the strength of each signal, the reasons and the limits. The weights, the amount each reason took off and the raw sum stay with us. Accounts are free to create, so anyone can point the widget at their own key. A page that said exactly how much each behavior counts would double as a manual for getting past the check. The signals, reasons and limits are yours; the arithmetic is not.

Older entries and availability

Entries from before 12 October 2026 are only partly recorded: the signals and reasons are there, while the limits and policy steps say “not logged”. From that day on, every evaluation records how it came about, so newer entries are complete. The page is available in every account and every plan, and there is nothing to switch on.

Log in to your dashboard, open an entry under User Interactions and look for the button.

Ready to ditch CAPTCHA?

Start protecting your forms — invisibly. No credit card required.