Talk to the hottest leads first.

AI lead scoring ranks every lead by how well they match your buyer and how ready they look, so your messages go to the people most likely to pay. In Gemlead, Dana AI does it for every lead Simon AI finds on X: she reads the profile, the posts and what brought them in, compares them with your persona, and gives one fit score from 0 to 100. At 60 and above a lead is hot, and Dana AI also tells you when to write.

What talking to the wrong leads costs

Your outreach capacity is small, and on X it is capped. An account can send a limited number of DMs a day (X limits), and a Gemlead campaign's daily leads limit starts at 75. Every message spent on someone who will never buy is one a buyer did not get.

The data shows what that waste costs:

• Reputation. 73% of B2B buyers actively avoid suppliers who send irrelevant outreach, in a Gartner survey of 632 buyers. In Hunter's State of Email Outreach 2026, 61% of decision makers cite irrelevance as a reason cold emails fail.
• Reply rates. In the same Hunter data, sequences sent to 21 to 50 people averaged 6.2% replies, against 2.4% for sequences sent to 500 people or more.
• Titles mislead. C-level executives are 30.2% less likely to reply to cold emails than other employees (Gong, 2026). A big title is not a hot lead.
• Time. Sellers spend only 40% of their time selling, according to Salesforce's State of Sales 2026, a survey of 4 050 sales professionals.

No public study measures X DMs on their own, so the reply figures come from cold email, the closest large dataset. They agree on one thing: a short list of the right people beats a long list of everyone. That is why "score before anything is sent" is a core rule in running an AI SDR.

How Dana AI scores your X leads

The first read: is this a buyer?

Every lead starts with a first read. Dana AI compares the person with your persona: its description, keywords, competitors and red flags. Do they buy what you sell? Have they already solved the problem? Below 60, that score is final. It costs 1 credit.

The second read: how hot, and when to write

A lead of a B2B persona that scores 60 or more gets a second read before it is marked scored. Dana AI reads their recent posts, how active they are, and how closely they resemble the people from this persona who already replied to your campaigns, up to about the last 50. A quiet account can drop inside the hot band, and a live buying moment can push the score up. A lead only falls below 60 if they turn out not to be the buyer, or already solved it. This read costs 1 more credit, and on B2B leads the judgment runs on JEV AI, a model built to judge instead of guess.

It also fills the Contact block on the lead:

• Reach them: the next window to write.
• Usually online: the hours they are typically on X.
• Their time: their timezone.
• Last post and Activity, in posts per day.

The window comes from when they actually post. The hour never changes the score: it only tells Nico AI when to send.

One read for B2C personas

When your persona sells to consumers, Dana AI reads each lead once: the profile, what brought them in (the post, the reply, the competitor they just followed) and their recent posts. 1 credit, hot or not. The rules fit consumers:

• Hot needs a buying sign. They joined or paid, with you or a competitor; they ask for the price, the link or how to join; they complain about what they use; they already use a competitor's offer closely. Without one, a lead stays under 60, however good the fit.
• A quiet account is not a bot. Few posts never lower the score.
• Sellers, brands and media score below 16, unless they show a buying sign of their own. Bots, giveaway farms, anyone under 18 and your red flags always do.

What you see

One score and a short reason on each lead. The status stays scoring until Dana AI is done, so you never act on a half-read. The hottest leads rise to the top, and the order keeps up as new leads come in. Then the best fits go to Nico AI's warm-up campaigns.

Why scoring before you send works

Most outreach fails at the list, before the first word is written. Scoring fixes the list:

• Small and relevant beats big. Lists of 21 to 50 people reply more than twice as often as lists of 500 or more, in Hunter's data above. Scoring gives you a short list without cutting your volume.
• The first vendor wins. Buyers purchase from the vendor they ranked first 77% of the time (6sense, 2025). Reaching the hot lead first matters more than reaching every lead.
• Timing counts. Companies that tried to reach a lead within an hour were nearly seven times as likely to qualify it as those that waited longer (Harvard Business Review). Leads found by a post or a reply are scored first, and the Contact window tells Nico AI when they are around.
• It learns from your replies. On B2B personas, each hot lead is compared with people who already replied. The longer your campaigns run, the sharper that signal gets.

Most sales teams already lean on AI for this. In Salesforce's 2026 survey, 87% of sales organizations use some form of AI for tasks like prospecting, forecasting, lead scoring or drafting emails. The rubric behind a good score, fit, intent and reachability, is laid out in how to find leads on X.

Scoring by hand vs with Dana AI

Who it is for

• B2B founders. Simon AI's Lookalike network can bring up to 1 000 leads an hour per account, more than anyone can read. Dana AI keeps the hot ones on top, so your campaigns start with the best fits.
• Agencies. One persona per client type, each with its own rules in the description, so each list is scored against its own buyer.
• Coaches and info products. Keyword search brings people describing the struggle. Scoring separates the ones who need help from the ones selling the same help.
• B2C sellers. A betting tipster's Competitor comments bring bettors and rival tipsters alike. Dana AI puts the bettor asking for the link at the top, and the rival tipster below 16.

How to get sharper scores

• Put your rules in the description. Everything Dana AI should know goes there, including the signals that raise a lead and the ones that sink it.
• Use red flags. List the profiles you never want, like students, job seekers or vendors who sell what you sell.
• One persona per buyer. Two customer types in one persona blur both scores.
• Set the market right. B2B and B2C leads are scored differently. Switching changes how new leads are scored; rescore a lead to apply it.
• Launch a campaign early. The look-alike signal needs replies to learn from.
• Pace your credits. A new persona's daily scoring limit starts at about 1/15 of your plan's monthly credits, about 3 300 on Starter and 10 000 on Business, spread across 24 hours.

FAQ