What it gets wrong
probr is a language model reading things people published. That is a genuinely useful thing to have pointed at your competitors, and it is not the same as a system of record. This page is the honest version, and it is short because there are only a few failure modes and they are all easy to state.
It misses things
A competitor can publish somewhere probr does not look — a regional trade publication, a closed community, a podcast, a conference talk with no writeup. It can also change something quietly on a page that probr read yesterday and will not read again for a day.
Coverage is good, not complete, and it is better for companies that publish in English and worse for ones that mostly do not.
What to do about it. Treat probr as the thing that catches most of it, not all of it. If a specific competitor matters enough that missing something would be serious, keep whatever you already do for that one. probr is meant to remove the work of watching everyone, not to be the last word on anyone.
It gets details wrong
The model can attribute an announcement to the wrong company, especially when two companies are named in one article. It can state a number with confidence that the source states more carefully. It can describe something as having happened when the source said it was planned.
This is not rare enough to ignore. It is uncommon enough that the feed is worth reading, and common enough that you should not quote a signal without opening the link.
What to do about it. Every signal carries its source URL. Before you act on one — repricing, a competitive slide, telling a customer something — open it. It takes a click, and the link is right there for exactly this reason.
It has no opinion worth trusting on its own
probr says how much something appears to change, and it is guessing. It does not know your roadmap, your pricing pressure, which of your deals this competitor is in, or what your team already decided last month. A signal marked critical might be irrelevant to you, and a low one might be the most important thing you read all month.
The importance rating is there to sort a feed, not to make a decision.
What to do about it. Read the direction and the impact as a starting sort order. The judgement is yours and it needs context probr does not have.
It is a day behind
Discovery runs once a day. Something announced at noon usually reaches your feed the next morning. Briefs come once a week.
If you need to know within minutes that a competitor changed their pricing page, probr is the wrong tool and no amount of configuration will fix that — it is a choice in the design, not a limit we are working around.
What it does not do at all
Worth saying plainly, because people ask:
- It does not measure competitors' traffic, keyword rankings, ad spend, or conversion. It reads what they publish.
- It does not access anything private. No logged-in areas, no paywalled articles it has no right to, no purchased data sets.
- It does not track individuals. It reads companies, and the people it mentions are mentioned because a company announced them.
Why we put this in the documentation
Because the alternative is that you find out on your own, at the worst moment, and stop trusting anything in the product.
A tool that reads for you is only worth having if you know where its reading stops. Everything above is a real property of how probr works, not a disclaimer — and the one habit that covers almost all of it is opening the link before you act.