Signado

A collection engine with a model layer doing the sorting, and the sorting is what the vendor sells. Finding the people in a LinkedIn conversation, enriching them and keeping their comment needs no model, and with scoring and drafting removed a buyer would still hold a working monitor of who engages with which posts.

What the model adds is the judgment a buyer would otherwise make by hand: a fit score from zero to 100 that weighs role, company and what the person said, a one sentence reason, a suggested next move, and a first draft. That sits closer to the center of the value than in most signal tools, because the promise is qualified leads rather than raw engagers.

The vendor also barely markets itself as an AI product, so this is a plain C for a product that would survive the removal test, not a marketed claim that fails it. Ask which parts of the workflow stop working with scoring and drafting switched off.

Human review is the architecture rather than a setting. The product never posts, comments or sends LinkedIn messages as the customer, drafts are presented as the customer's to edit, and nothing reaches a prospect until a person sends it from their own mailbox or pushes it to a sending tool. Spend has real governors too.

Each plan carries a daily discovery budget the customer can change, a source can run once, daily or on chosen weekdays, discovery pauses rather than overspends when credits run out, and work started through the protocol server shows its projected credit cost and needs confirming first, with every paid call reporting what it cost and the balance left. Held below the top band because these controls govern spend and timing rather than content.

Nothing describes guardrails on what a draft may say, and once a lead is pushed to a connected sequencer the platform's oversight ends. Ask whether pushes to sending tools can run automatically, and what review sits between a draft and that push.

The scoring is explained one lead at a time, and the models behind it are never named. Every score arrives with a short written reason, and the methodology page says what goes into it: role fit, company fit and the substance of the comment, combined into one number. That per record explanation is more than most signal tools give, because a user can see why a person was ranked and overrule it. The layer underneath is missing.

No model provider or family is named for scoring, company research or drafting, no accuracy or precision figure is published for the fit score, and the terms call scores estimates with no guarantee of predictive value. Posts are read in any language and drafts are written in English, Spanish, French or German, with nothing on how scoring holds up outside English. Ask which models score and draft, and how often a strong fit turns out to be a real buyer.

One measured result with its working shown, which is rarer in this index than a bigger claim. The vendor reports that LinkedIn messages to people who had just posted or commented drew a 52 percent reply rate, against 18 percent for messages to people from a contact list, and publishes the counts behind both: 78 replies from 149 messages and 25 from 139, across 288 messages over two months. The arithmetic reconciles.

Reviews on three public platforms add first name testimony, including an agency owner reporting ten meetings booked in a month from leads the product found. Held below the top band because the measurement tests the method more than the product at scale: it is the vendor's own outreach, the sample is small, and no result across a customer cohort is published. The terms add that marketing figures come from select experiences and are not guaranteed. Ask whether the comparison has been repeated across customer accounts, and what share of strong fits lead to a reply.

Every step up to the send happens inside the product, and responsibility for what follows is assigned to the customer in plain words. The platform finds a person, appends an email address they never supplied, drafts the message and can send it from the customer's Gmail or SMTP account or push it into a sequencer. The terms then make obeying CAN-SPAM, GDPR and other outreach rules solely the user's job, prohibit unlawful bulk email, and require users to honor opt out and deletion requests.

That is a clear allocation, not a posture. The one control that can keep a person out of discovery is an exclusion list for companies and contacts, and its size is set by price: ten entries on the entry plan, three hundred on the top one, held separately in each workspace. Nothing describes how an objection made to one customer reaches another, or what consent basis supports emailing someone on the strength of a public comment. Ask what suppression exists beyond the capped exclusion list, and how a person who objects is excluded everywhere.

More published than most young vendors, with one contradiction that sits exactly where it matters. The privacy policy names every subprocessor by service, from sign in and payments to the database and email, says data is stored in the United States or Germany, and the terms cite the 2021 Standard Contractual Clauses for transfers out of Europe.

A processing agreement is offered on request to paying customers, and people whose details sit in the database are told how to ask for access or deletion. The problem is the role the vendor takes for those people. The privacy policy says it handles scraped contact data as a processor on the customer's instructions, while the terms say it acts as a joint controller, and under GDPR those are different roles with different duties.

Either way the lawful basis for collecting the data is placed on the customer, although the collection runs on the vendor's infrastructure. Ask for the processing agreement, and which role the vendor accepts for the people it monitors.

Provenance at the level of each lead is built in, and the source is named, which several comparable tools avoid. Every record keeps the post or comment that produced it, so a buyer can check the public evidence rather than trust a score, and the vendor names LinkedIn openly while stating it has no affiliation with it.

The signal types are listed: post authors, commenters and reactors from daily discovery, plus funding, hiring, leadership changes, launches, partnerships and news when a customer runs the deeper company research. Held below the top band on the layers beneath the signal. The terms say data is aggregated from various third party sources without naming any, the supplier behind found email addresses is never identified, and the method used to collect LinkedIn activity is undescribed. A buyer can see what was said, but not how it was gathered or where the contact details came from. Ask which providers supply email and company data, and under what license.

The buyer's own account is out of the picture, and the platform question moves to the vendor. No LinkedIn login, cookies or extension are requested, and the product does not post, comment or message as the customer, so the account restrictions that define the automation tools in this lane cannot reach the buyer. That is a real architectural difference and the vendor makes it the center of its pitch.

What goes unexplained is how activity is collected from LinkedIn on the vendor's own infrastructure, and under whose terms. The terms then make no representation about third party platform rules, note that platforms may restrict automated collection, and ask the user to comply with those rules, although the user has no say in how collection is done. What the buyer inherits is continuity risk in a service whose only input depends on access LinkedIn controls. Ask how engagement data is collected, and what happens to the service and to stored leads if that access is withdrawn.

A plain promise on training, made in public and broad in scope. The vendor states that it does not sell customer data or use it to train AI models, which covers models generally rather than only public ones, and that leads sit in a private workspace with access controls. Access through the protocol server is scoped to one approved workspace and the same row level security applies to every call, which is a real boundary between customers.

Held below the top band because the promise lives on the marketing site rather than in the terms or privacy policy, and because the model providers that process posts, profiles and drafts are never named, so a buyer cannot check what those providers keep. The statement also leaves open whether it covers the monitored people's posts and profiles as well as the customer's own lists. Ask which providers process lead data and drafts, what they retain, and whether the no training commitment appears in the processing agreement.

The design tells a prospect why they were contacted, and nothing tells them how the message was written. Every lead starts from something the person chose to post publicly, and the drafted opener is built to reference that comment, so the recipient learns exactly what surfaced them. That is a genuine authenticity property and the opposite of tools that sell undetectable outreach. The other half of the axis is silent.

Drafts are written by a model, and nothing published takes a position on whether a recipient is told so, or on Article 50 of the EU AI Act, although the vendor is established in the EU. People found in conversations are not notified that they were collected, and the only route offered to them is an email address for privacy requests. Ask whether drafts are marked as generated when they reach a recipient, and what the vendor considers its obligation under Article 50.

Deep where the product meets AI tools and sending tools, thin where it meets a system of record. The protocol server is documented more fully than most in this index: an endpoint, connection steps for six named clients, browser sign in with no keys to manage, and a table of every tool with what it does and what it costs, from listing and exporting leads to managing sources, finding emails and drafting.

Leads push natively to Instantly and HeyReach, both indexed here, send from Gmail or any SMTP account, and reach Zapier, Make, n8n and webhooks, with CSV export on every plan. Held below the top band because no native CRM connector is listed, so getting a qualified lead into Salesforce or HubSpot means building a webhook flow, and no conventional interface is documented beyond the protocol server. Ask whether CRM connectors are planned, and whether an interface exists outside the protocol server.

A hosted service with its residency stated, if not chosen. The terms say the service is hosted in the United States and that European users' data moves there under Standard Contractual Clauses, the privacy policy says data is stored or processed in the United States or Germany, and the hosting, database and automation providers are named. A buyer can tell a regulator where the data sits and through which companies, which beats the silence common in this lane.

Two things hold it below the top band. The two documents describe residency slightly differently, and the privacy policy adds wherever partners and providers operate, which leaves the edge open. And there is no regional choice, so a European customer of a French vendor has its data hosted in America by default. Ask which records are held in Germany and which in the United States, and whether European hosting is available.

Basic controls asserted, nothing attested. The vendor states that workspaces are private with access controls, that connections are encrypted, and that the protocol server applies row level security to every call and uses browser sign in rather than keys a customer could leak. Those are sensible design choices. No certification, audit, penetration test, trust page or security contact was located, and the privacy policy's reference to industry standard measures describes nothing specific.

The exposure is lower than for the LinkedIn automation tools in this lane, because no LinkedIn session is ever held. Even so, customers who send from Gmail connect a mailbox, and the platform stores enriched profiles of people who are not its customers. A missing audit is expected for a single founder business, and it is still what a buyer meets. Ask what security review has been done, and what access the Gmail connection grants.

Everything a buyer needs to budget their own purchase is on the page, including the number most signal vendors leave out. Four self serve plans are priced at $29, $59, $99 and $349 a month, with credits, lead ceilings, source slots, contact storage and daily discovery budgets stated for each, and no seat fees.

The burn rate is published per action: five credits to find, enrich and score a new lead, one when the lead turns out to be on the exclusion list, five for an email and only when one is found, three for a draft, forty for deeper company research, and nothing to view or store.

Extra workspaces on the top plan are priced at $349 each, discovery pauses rather than overspends when credits run out, and the trial and refund terms are specific, including that the seven day trial needs a card and converts on its own. Two loose ends keep this from being perfect. Credit packs are described but not priced, and the pricing page says purchased credits never expire while the terms say they lapse twelve months after the last active subscription. Ask what a credit pack costs, and which expiry rule applies.

Leaving is cheap and export is free, while the terms around termination lean the other way. Plans are monthly, cancellation is self serve from settings and takes effect at the end of the billing period, and leads export as CSV in the app or through a free export tool on the protocol server for any date window, so a customer can keep a running copy rather than depend on a handover. The terms are less reassuring.

They reserve the right to delete an account and its content at any time without warning, make the customer solely responsible for its data, waive claims for data loss, and say nothing about how long records are kept after cancellation. Contact storage is also capped by plan, from 3,000 records on the two smaller plans to 18,000 on the largest, which a long running customer should watch. Ask what happens to leads, sources and drafts after cancellation, and whether notice is given before an account is removed.

The product touches sending at two points and governs neither in public. Customers can send drafts straight from their own Gmail or SMTP account, and can push leads into Instantly or HeyReach, which carry their own sending controls. Two things work in a sender's favor. Emails are charged only when one is found, which suggests some verification, and the rule of contacting only people with a real reason keeps volume low by design.

Beyond that the record is empty: how addresses are verified, whether catch all or role addresses are filtered, what limits apply to Gmail and SMTP sends made from inside the product, and how bounces and complaints are handled are all unstated. Ask how found emails are verified, and what limits govern sends made from the platform itself.

Who the product is for is stated precisely, along with when it works. The vendor names its buyers as AI automation agencies, SEO, paid media and content agencies, go to market and marketing consultants including fractional leaders, and small B2B teams selling those services, with separate pages for agencies, consultants and teams.

It also names the condition that decides fit: the product works when a customer's prospects post or comment on LinkedIn, which tells a buyer whose market is quiet there not to buy. Language coverage is stated as well, with posts read in any language and drafts written in four. The single platform focus is the boundary to weigh, since a market that lives on other channels is out of reach by design. Ask how lead volume varies by market, so a buyer can judge whether their niche is active enough.