Working with leads

Feed: views and filters

The Feed is your main working screen: everything the campaigns find lands here, from all channels at once. One card = one signal; the list is on the left, the details of the selected card on the right. Views The switcher above the list: - All - everything the campaigns collected. - AI-selected - signals the AI recognised as leads; this is where the main work happens. - AI-filtered - the AI looked at them and rejected them. The list is not hidden forever: if the AI got it wrong, the card has a "This is a lead - restore" button and the signal goes back into your queue. - Archive - everything already handled. Filters - Campaign - show the signals of a single campaign (all campaigns by default). - Unread only - a quick way to see what is new since your last visit. Filters combine with views; if nothing matches the combination, the screen says so and offers to reset the filters. What is on a card - who wrote it and where (channel, chat or post, campaign); - the text of the message itself; - the AI verdict with a score - why this is a lead (expands on click); - actions: reply draft, sending, "Not a lead", a reminder task. For the details on these actions see the articles "Working with a signal" and "Lead actions" next to this one in the same section. Where the cards come from The Feed is shared by all campaigns and channels: Telegram chats, Threads search, LinkedIn, Reddit, X, Bluesky, Hacker News, replies to your own posts. In parallel, every signal is carried out by the campaign deliverers - to a Telegram bot, Slack, email or your CRM via webhook - so you can work without even opening the app (see the article "Multiple lead delivery channels and their order").

Replying to a lead

You write the reply in the signal's detail pane, in the field pinned at the bottom. It is a single field that grows with the text. If the responder agent has already prepared a text, the field is called Reply draft and holds a ready version. If there is no agent, the field is empty and called Write a reply manually. Sending The buttons below the field depend on the channel: | Channel | Buttons | | --- | --- | | Telegram | Send as DM and Send to chat | | Threads | Send to thread - a public comment under the post | | LinkedIn, Reddit, X, Bluesky, Hacker News | Copy comment - the text goes to the clipboard and you publish it | ⚠️ Important: direct sending is not available on LinkedIn, Reddit, X, Bluesky and Hacker News yet - there we always prepare a draft that you copy and publish yourself. If your channel account is not connected or the connection has expired, Telegrin shows a banner about it with a button that takes you to settings, and offers the text for copying. Refining the text There is a row of tools above the field. Four quick buttons apply in one click: - Shorter - Friendlier - Add a question - Drop the pitch Next to them is the What should we change in the reply? field. Write your instruction in plain words ("mention that we work in Kyiv", "drop the mention of price") and press Enter - AI rewrites the text with your correction. If the new version is worse, press restore the previous version - as many steps back as you like. ⚠️ Important: every refine costs 1 credit - the same as a full reply generation. That applies both to the quick buttons and to a correction sent with Enter. If you only need to change a couple of words, editing the text by hand right in the field is cheaper and faster. What a reply costs | Action | Price | | --- | --- | | Editing the text by hand in the field | free | | Sending the reply | free | | Generating or refining the text with AI | 1 credit | | A Lead / Not a lead decision, restoring from the archive | free | If the message did not go through Telegrin shows a human-readable reason and returns the text to the field - nothing is lost. The most common reasons: - the lead's privacy settings do not allow a first message; - the lead has blocked your account; - Telegram has temporarily limited the sending rate; - the message is longer than 4096 characters - Telegram will not accept that much; - the account connection dropped. The card gets a Reply not sent marker: you can try again from the same pane. If the agent sent the reply itself When automatic sending is on for the responder, there is no reply box - instead you see a line saying the agent completed the step automatically. The text that was sent stays visible in the conversation. Continuing the conversation after the first reply After the reply, the conversation continues in the same pane: write in the field and send it like a normal message. The signal's verdict does not change, and the exchange is stored on the lead's card. How to make replies better If the texts are systematically off, the problem is almost never the individual reply: 1. The company brain. The most common cause of generic replies is empty or overly short Product and Advantages sections. AI does not invent facts about your business: what is not in the brain will not be in the reply. 2. The responder prompt. A prompt is an instruction, not a template: the agent writes its own short text every time (usually 2-4 sentences with one call to action). So it paraphrases and compresses, and can drop a specific wording. 3. To make sure a piece of text ends up in the reply, write it in the prompt as an imperative: "always add the link example.com at the end, word for word" works, "we have a website example.com" does not. Descriptive sentences are treated by AI as background it may ignore. 4. Give an example of the reply you want right in the prompt, in quotation marks - one or two samples work better than an abstract description of tone. ℹ️ Note: the language of the reply is not set by the prompt. The agent replies in the language of the post or of the lead's message - that rule outranks both the prompt and the brain. All you can change is the fallback for cases where the language could not be determined.

AI-filtered signals and the archive

Not everything Telegrin finds reaches your feed as a lead. Some of it is filtered out by AI, some of it you archive yourself. Both piles stay available - nothing disappears without a trace. AI-filtered The AI-filtered view in the feed shows the messages AI checked and judged to be off-target. A credit was spent on that check, so it can be useful to see what exactly you paid for. Next to each card you can see the AI verdict and a short explanation of the decision. This is the best tuning tool you have It is worth looking into AI-filtered regularly during a campaign's first days: - You see leads you wanted there - AI is too strict. Lower the Score threshold on the analyst agent or refine its prompt. - You see solid noise - everything is working correctly. If there is a lot of noise, narrow the keywords or add stop words: that way you stop paying to check the obviously off-target. - You see the same kind of noise repeatedly - add a word typical of it to your stop words, and next time it will be filtered out for free. Bringing a signal back If AI got it wrong, press This is a lead - restore. The signal returns to the queue and waits for your decision. ⚠️ Important: you cannot reply straight from the AI-filtered view - first bring the signal back with the This is a lead - restore button, and then the reply field appears. The This is a lead - restore button is shown on the card only in the AI-filtered view, but in the detail pane it is always available. Restoring a signal does not spend credits and does not send anything. The archive Everything you mark with the Not a lead button goes to the archive. Right after you press it, an Undo bar appears at the bottom - about seven seconds to change your mind. While it is up, the decision can still be taken back completely. Archived cards sit in the Archived view and look like a collapsed row with an Archived marker. You can bring a lead back into play with the Lead button right on that row. ⚠️ Important: archiving does not recall a reply that was already sent. If the message went out, it was delivered - archiving only removes the card from your working list. When a prospect replies to an archived card It happens that you archive a signal and the person writes a day later. That conversation comes back by itself into the working feed with a Prospect replied marker - and you can answer it right away from the usual field. This is deliberate: a live conversation matters more than your earlier decision, and Telegrin will not let it get lost. How the archive differs from blocking | | Archive | Blocklist | | --- | --- | --- | | What it does | Removes one specific signal from the list | Stops all future signals from that person | | Scope | One card | The whole workspace, all campaigns | | New signals from this person | Keep appearing | Are not created | | How to reverse it | The Lead button on the archived row | Settings → Blocklist | 💡 Tip: archive what simply is not a fit right now. Block only spammers and people who asked you not to write again - blocking applies to every campaign at once.

Pipeline: lead stages

The Pipeline shows all your leads by stage - from first contact to deal. If the feed is "work through the inbox", the pipeline is "what I have in progress right now". Four stages | Stage | Who lands here | | --- | --- | | New | The lead has just appeared, nobody has worked on it yet | | In progress | A message has been sent to the lead, the conversation has started | | Qualified | The lead replied, the interest is confirmed by the exchange | | Rejected | The lead was judged to be off-target | Stages move on their own This is the main difference from an ordinary CRM board: you do not need to drag cards. The stage changes based on what the agents and you do - for example, the moment a message goes to a lead, it moves to In progress. You can still change the stage by hand, when the real state of the negotiation differs: - open the lead's card and use the Stage selector at the bottom of the pane; - or select several leads and press Change stage. Two views The switch at the top: Kanban (columns) or Table. The table shows the lead, the channel, the campaign, the stage and the update date, and you can sort by clicking a column header. Filters - Channel - Telegram, Threads, LinkedIn, Reddit, X, Bluesky, Hacker News or all; - Campaign - one specific campaign or all; - Period - 7 days, 30 days or all time; - Search - by name or text. The lead card Clicking a card opens the pane: - Conversation - the full exchange with the lead; your messages are labelled "You"; - Agent actions - what the agents actually did with this lead: how the analyst scored it, whether the responder sent a reply automatically, whether a draft is waiting for approval; - Write - send a message straight from here, without going back to the feed. The button is only available on channels we can send to (Telegram and Threads), and Telegrin states where exactly the text will go; - Stage - a manual stage change. Limits The board shows up to 500 leads at a time, and the filters work on the loaded selection. If you have more leads, narrow the period - that way you see the current picture. The pipeline does not refresh itself: to see fresh data, reload the page. How the pipeline connects to the feed Leads reach the pipeline from the feed - every signal you accept with the Lead button creates or updates a lead on the board. So the working order is this: you work through the inbox in the Feed, and you follow the state of your deals in the Pipeline. 💡 Tip: the Qualified column is the most useful indicator of how the system is doing. If it is empty while In progress is full, the issue is usually not the number of leads but the text of your replies: people receive them but do not answer. Review the company brain and the responder prompt.

Home: metrics and analytics

Home answers one question: what your campaigns produced over the chosen period and what it cost. There are two filters at the top: the channel switcher (all or one specific channel) and the period - 7, 14 or 30 days. Every number on the page is calculated for that selection. Four metrics | Tile | What it means | | --- | --- | | New signals | How many messages became signals in the period; what AI filtered out is not counted here | | Leads | How many leads you approved in the feed | | Credits spent | How many credits went out in the period, and how many are left on the balance | | Filtered by AI | What share of off-target messages AI filtered out for you | ℹ️ Note: if a metric cannot be calculated for the selected channel, "no data" is shown instead of a number - that is neither a zero nor an error. For example, Filtered by AI is measured for Telegram only. Monitoring funnel It shows a message's path through the stages: Messages → Keyword matches → Checked by AI → Filtered by AI → Signals. This is the most useful block for tuning. Read it like this: - a big gap between "Messages" and "Keyword matches" - the keywords are too narrow, or the chats are the wrong ones; - a big gap between "Checked by AI" and "Signals" - AI checks a lot and keeps little. You are paying to check noise: add stop words or narrow the keywords; - "Filtered by AI" is close to zero - the filter is too soft, everything is flying into the feed. Raise the analyst's score threshold. The funnel is calculated for Telegram only, so it is not shown when an external channel is selected. Top Telegram chats A breakdown by chat: how many signals each one produced, how many of them got a reply and what the conversion is. 💡 Tip: a chat that has produced no signals for months is worth removing from the campaign - it adds load and noise but no result. And the other way round: if one chat produces most of your leads, look for similar communities. Top keywords The same thing, but for LinkedIn, Reddit, Threads, X, Bluesky and Hacker News: which queries actually bring signals and replies. Pick the relevant channel in the switcher at the top to see this block. Keywords that never find anything A separate block at the bottom of the page - keywords that have not produced a single signal, ever, even though the campaign keeps checking them. For each keyword you see the evidence: how many days the campaign has been scanning it and how many posts were checked. The Remove button takes the keyword out of the campaign. ⚠️ Important: on external channels every such keyword is real money: the search for it runs every time and can cost credits even with no finds. Review this list at least once a week. A keyword is not suggested for removal until the campaign has been running long enough - the conclusion is not drawn from a single day. If a keyword is the only one in a campaign, you will not be able to remove it: without it the campaign has nothing to search for. How to read the overall picture 1. Few signals - look at the funnel: is the problem at the keyword stage (few matches) or at the AI stage (a lot filtered out)? 2. Many signals, few leads - AI is letting off-target items through. Refine the analyst prompt or raise the threshold. 3. Many leads, few replies - the bottleneck is not in the search but in your process: signals are piling up unprocessed in the feed. 4. Credits melting fast - look at Credits spent per channel and check the block of keywords that never find anything.