Signal reads the public web for your company, your competitors and your market. Every post is read by a model that decides whether it is really about you and whether someone should act on it — and shows you the words it decided on. An agent sets it all up from your website.
Northwind’s CSV export has been broken since Tuesday’s update and support hasn’t answered my ticket in four days. I need that file for a client report on Friday. Has anyone found a workaround?
WhyNames Northwind’s export, says an update broke it, and has a deadline with no reply from support.
Why it’s a lead: names a competitor, a renewal date and a team of forty.
Of the posts judged to be about one of the four.
An example night. Your own numbers are printed on every brief, so you can always see how much was read to reach them.
No keyword lists to write, no boolean queries to maintain. The agent does the setup and the reading; people decide what to do about what it found.
Your names and products, the competitors a buyer weighs you against, the topics your market argues about, who buys and why — and what else shares your name.
Every proposal is a chip you can edit, with the reason it was proposed and where it was found. Nothing is read or spent until you confirm.
The last 30 days on the first run, then every day. What needs someone comes first, with the post, the sentence and how sure it is.
Split by sentiment, because raw volume rewards a crisis.
Why it’s a lead: asks for a recommendation, gives a team size, and has nothing in place today.
Share of voice, how a launch landed, and every time someone weighs you against a competitor.
Feature requests and what broke, in the words of the people it broke for.
Public questions and complaints, found before they become the thread everyone reads.
People asking for what you sell, and a competitor’s customers who are unhappy with it.
A weekly brief: what changed, what needs a decision, and the posts behind each line.
Keyword alerts match all of them, and a dashboard built on keyword alerts counts all of them. That is why most listening tools are mostly noise.
Signal deals with it twice. When it sets up, the agent tells you what it expects your names to collide with, before a single post is read. After that every post is read against what your company actually is, and anything that is not about you stays out of your counts, your share of voice and your brief. It is not deleted: you can see what was ruled out, and why.
Just saw Meridian on opening night. That last scene. Still thinking about it.
WhyAbout a film called Meridian, not the product. Nothing in it touches software or Northwind.
Is Meridian’s API rate limit per workspace or per key? The docs don’t say and we’re about to roll it out to three teams.
WhyAsks about the API of Northwind’s Meridian product, ahead of a rollout.
When a source fails or refuses us on a run, the run says so. An empty result never stands in for a broken one.
A model reads for you, so you should be able to see what it read, how sure it was, and what it cost. You can, everywhere.
Replies are drafts. You send them yourself, from your own account, or not at all.
A sentiment, a kind of post, a line in the brief: each links to the posts it came from.
Every reading carries a confidence. Low ones are shown as unsure and kept out of the counts, never passed off as fact.
A trial runs on a fixed $10 budget. The agent checks it before every paid call, stops at the cap, and shows you what has been used.
It will not set up monitoring of a named private individual, whoever asks.
Only your workspace’s members see it. Export your leads whenever you like; we delete a workspace when you ask.
When it sets up, the agent writes a description of your company from your own site and lists the namesakes it expects. Each post is then read against that description by a model, which answers yes, no or unsure, with a confidence.
You can read its reason on every post, and change the description whenever you like.
No. It drafts a reply when you ask for one. Sending is always a person, from their own account.
Start with a free trial: 14 days on a fixed $10 budget, no card. The budget is shown in money and the product stops at it.
After the trial, talk to us. The price depends on how much you listen to.
A few minutes. Paste your website, confirm your email, check what the agent proposes, and say go. The first run reads the last 30 days; after that it reads every day.
It is a model reading posts, so it is sometimes wrong — and it says how sure it is, every time. Readings below a confidence threshold are shown as unsure instead of being counted.
When it gets one wrong you can see exactly what it read and why it decided, which is what you need to correct the description it works from.
The model reads posts in most languages. Searches use the names and terms you confirm, so add the ones people use in each of your markets.
LinkedIn forbids automated collection in its terms and has enforced that in court. We would rather tell you now than have you find the gap later.
The members of your workspace, and nobody else. Workspaces never share data with each other.
Paste your website and confirm your email. The agent reads your site and shows you what it would listen for, and nothing runs until you say so.