Building a Targeted Campaign That Converts

That changes what "targeting" means. You are no longer assembling a static list from filters. You are setting two things: who qualifies, and what they have to be doing for you to reach out. This guide walks through both.

Everything downstream depends on this, so get it right before you touch anything else.

You need clarity on:

In Alsona this lives in your account targeting, set during onboarding and editable afterward. When you created your account, the platform scanned your website and pre-filled your industries, job titles, company types, and company sizes. Check those before you build an agent, because every new agent inherits them.

Two things to watch. Target locations is usually the one field the scan leaves empty, so fill it in yourself. And every targeting dropdown has include and exclude toggles per row, which most people miss. Excluding is often more powerful than including. If you sell to marketing teams, excluding agencies can lift reply rate more than any amount of extra include criteria.

Pro tip: if you already have customers you are happy with, reverse engineer your ICP from them rather than from who you wish you sold to.

This is the step that replaces list building, and it is where most of the leverage is.

Alsona tracks 31 signal sources across nine categories: hiring and workforce, digital and social, ads and demand gen, reviews, news and media, events and talks, tech and dev intel, financial, and alternative data. Each one is a different kind of evidence that a company is in motion.

You are answering one question: what would a company be doing right now that makes this a good week to talk to them?

A few examples of how that maps:

Start with four to six sources, not twenty. Every source you enable is another stream of detections, and detections are what your plan meters. An agent watching six well-chosen sources almost always beats one watching everything, because the signal-to-noise ratio is what determines your reply rate.

Once you toggle a source on, a Customize dropdown appears with Track chips that tighten what qualifies. Most people never open it, and it is the single highest-value control on the page.

LinkedIn audience engagement requires keywords. It will show a warning icon until you add them. Without keywords it would surface everyone who reacted to anything your target companies posted, which is noise. The keywords are what turn it into a real signal.

The old version of this article suggested using third-party data tools to find technologies in use and company growth signals. You do not need them. Technology fit and technology stack changes cover the tech stack, and hiring signals plus SEC filing intelligence cover growth and funding. Keeping that research inside Alsona means it feeds the agent directly rather than sitting in a spreadsheet.

The old workflow was to review every prospect by hand and cut the ones who did not fit. Opportunity score does that automatically, and it is off by default, so switch it on.

In agent settings, turn on "Calculate opportunity score for agent prospects." You then get a balance bar between ICP fit and intent, defaulting to 60% fit and 40% intent. Move it toward fit if you have a narrow market and a broad set of signals. Move it toward intent if your ICP is wide and timing is what matters.

Two separate thresholds control what happens next, and they sit on different screens:

Both offer 25+, 50+, 75+, and 90+. Both do nothing until scoring is switched on, which catches people out.

Start at 75+ for outreach. If volume is too low, drop to 50+ and watch what happens to reply rate. If quality is inconsistent, go to 90+ and accept fewer conversations. The visibility threshold can sit lower than the outreach threshold, which lets you see who is being held back and decide whether the bar is right.

You still have manual controls where you want them. The Prospects tab shows everyone the agent found along with their score, so you can review before or during a run. The blacklist excludes by name, profile URL, job title, company, website, or email, with exact, contains, starts with, and ends with conditions. Use it for existing customers, current pipeline, and anyone you have already burned.

A signal-triggered message that ignores the signal wastes the whole system. This is where targeted agents win or lose.

Step 2 of the agent wizard, Pitch and personality, sets your offer, the pain points you solve, your conversation goal from a list of 15, tone (professional, friendly, or bold), and message length. These pre-fill from your account and are worth adjusting per agent, because an agent built on hiring signals should not sound identical to one built on review monitoring.

In the workflow step editor you choose between writing your own message and AI contextual outreach. Contextual outreach writes each message against the specific prospect and the signal that surfaced them. Your own message gives you exact control. Message variants let you run A/B copy inside one step, and there is a Help me write button if you want a starting point.

Guidance that holds up regardless of which you choose:

Then pick your outreach mode. Full autopilot sends and handles replies. Outbound autopilot sends and hands replies to you. Co-pilot drafts everything and waits for your approval. If you are new to the platform or the client is sensitive, start on co-pilot for a week, read what the agent writes, and move up once you trust it.

The Intent signals tab gives you a four-stage funnel: signals detected, high-intent companies, identified prospects, and pushed to outreach. Read it as a diagnostic.

Below the funnel, the breakdown by signal source shows detected, contacted, reply rate, and positive replies for each source you have on. This is where you find out which of your six bets was right. A source with high detection volume and a weak reply rate is costing you signal allowance and producing nothing. Narrow it or turn it off.

The workflow canvas has per-step counters showing how many prospects passed and failed at each point, so you can see exactly where a sequence stalls.

Change one thing at a time and give it long enough to mean something. Swapping three signal sources and rewriting your opener in the same week tells you nothing about which one moved the number.

Signal search is the default for a reason, but the other prospect sources still exist and still have their uses.

The rule of thumb: use a list when you already know exactly who you want to reach and the timing does not matter. Use signals when you know what kind of company you want but not which ones are ready this month.