When I find an unfamiliar product, the feature list is rarely the missing piece. Landing pages are good at telling me what a product wants to be. They are less reliable at telling me where a claim came from, whether it is still current, what the interface actually looks like, or what has not been tested.

Those questions determine whether I keep reading. I now look for four signals before I trust a product page enough to spend more time on it.

1. Original sources

Different sources answer different questions.

An official website can show current positioning. An app-store listing can establish platform availability, the publisher, and declared privacy information. Documentation can explain permissions and data flow. A public product interface can show what is actually available without an account. A maker's reply can clarify intent, but it is still a first-party statement rather than an independent audit.

Search snippets and directory descriptions are useful for finding leads. They should not silently become evidence. A useful profile links back to the page that supports each important claim and makes the source type visible.

2. A verification date

Product information expires.

Pricing changes. A free tier disappears. An extension adds a permission. A product moves from a waitlist to a working release, or from an active domain to an abandoned one. A statement that was accurate six months ago can be misleading today without anyone intentionally writing something false.

A visible verification date gives the reader a simple but important piece of context: this is when someone last checked. It also makes future corrections easier because there is a known baseline to revisit.

3. Real product screenshots

A logo proves that a brand asset exists. It does not prove that a product has a usable interface.

Real screenshots can answer practical questions immediately: Is there a working surface? What is the main action? Which controls are visible? Does the current interface match the written description?

The source still matters. A publisher-provided screenshot is evidence of what the publisher chose to show. A screenshot captured during a limited public check is evidence of what was visible in that check. Neither should be described as a full hands-on test if no account was created and no core workflow was completed.

Decorative mockups and generated illustrations can be good design assets, but they should not replace product evidence.

4. Explicit boundaries

"Worth a look" is not the same as "safe for everyone."

Reading public pages is not a security audit. Opening a demo is not the same as installing an extension. Reviewing a privacy policy does not prove that every data path behaves exactly as described. A product can be promising while still leaving important questions unanswered.

The most useful profiles say both what was checked and what was not. They also explain who may find the product useful, who may need stronger evidence, and which claims still depend on the maker's own documentation.

A correction should improve precision, not inflate confidence

Aye is a useful example. After Oka Apps replied to a factual review request, the profile was updated to include Windows availability, confirm that the displayed images were official, and record the maker's statements about Kimi 2.5 and user approval for sensitive actions.

Those clarifications made the page more accurate. They did not turn first-party statements into an independent network audit. The profile still labels what came from the maker and keeps the unresolved data-flow and approval-coverage questions visible. That distinction is the point of the four signals: new evidence should narrow uncertainty, not hide it.

Putting the four signals together

My quick scan now looks like this:

1. Can I reach the original product, documentation, store listing, or repository?

2. When was the information last checked?

3. Do the images show the actual product, and is their source clear?

4. Which important claims remain untested or first-party only?

This does not replace trying the product. It helps decide which products are worth trying and what to verify next.

Where BitShovel fits

BitShovel currently presents 32 product and application profiles in English and Chinese. Each page focuses on the problem, current status, original sources, recent verification date, reasons to keep reading, and unresolved boundaries. Product images favor official or publisher-provided interfaces over generic placeholders.

The site is free to browse, requires no account, and has no advertising, affiliate links, paid rankings, or tutorials. It offers RSS, anonymous feedback, and product recommendations.

Browse BitShovel in English

If you try it, pick one product you have never seen before. I would especially value two pieces of feedback: which claim felt best supported, and which claim still needed better evidence?