What makes outbound work: volume ✕ targeting ✕ messaging

Harmonic has a golden equation for the effectiveness of a growth analyst: volume times targeting times messaging (how much outreach goes out, whether it lands with the right people, and whether the message earns a reply). Because the terms multiply, an analyst who skips the research to raise the volume sends twice as many emails for the same number of replies.

Each account on Susheel’s list is a large organization with thousands of employees spread across dozens of teams, only some of whom work with startups at all. Nothing on the list says which ones do, which account is worth approaching this week, or what the people on the right teams are working on. Every one of those answers can be found by hand, but finding them across the entire list would take more hours than there are in a week.

‍

Determining which account to work today

Susheel’s day starts in Scout, Harmonic’s AI research agent. Scout takes its instructions in plain language, and reasons across Harmonic’s company and people data, the user’s own lists and network data, and the live web. A scheduled Scout runs that research on a repeating cadence: describe the task once, set how often it should run, and Scout generates a summary when it finds something new.

Susheel built two scheduled Scouts by talking through what he wanted to see, then setting that research to repeat on a schedule. Both Scouts watch for signs that a team works with startups: a partnership, an investment, an acquisition, or a product built on startup data. He scoped them to his team’s target list to begin with, and has since widened them to surface companies nobody at Harmonic has listed at all. “They’re passively monitoring the corporate landscape for signals that an organization might want to use Harmonic,” he explains, “and they’re flagging those signals well ahead of the general-purpose models. So I don’t spend time watching.” 

Signals alone would leave Susheel where he started, with names and no order to them. “I ask Scout to surface the signals and weigh them based on whether they’re recent, whether they’re high intent, whether they’re partnership announcements, and score them,” Susheel explains, “so I know the best targets to reach out to for the day.” Scout wrote the scoring model itself: points by event type, with a fund launch or an exit worth more than a partner hire, and a multiplier that triples anything from the last thirty days. The flat list becomes a ranked one that reorders itself as new signals land. 

‍

Surfacing six names inside a company of thousands

The signal that surfaces an account tells Susheel the organization is working with startups, but it doesn’t tell him which team is doing that work, or who’s on it. For that, he runs a skill in Scout: a set of instructions specified once and triggered thereafter with a slash command. Susheel built his in his first weeks at Harmonic and named it “CVC People Finder” (for corporate venture capital).

“I’ve told it to identify every single Harmonic buying center within that company,” he explains. “The skill looks at the CVC team, the M&A team, strategy, innovation, partnerships, AI strategy, the accelerator, every possible team in there. Then within each team it gives me a list of people, where they’re located, and why they’d be a fit.” Each person comes back with a high, medium, or low fit score. The skill drops the teams whose work doesn’t touch startups, and it checks regional headquarters alongside the main office, so a venture arm run out of Singapore turns up with the one at home. At Airwallex, the skill returned a startups team and a corporate development team, and Susheel booked the head of each. The opportunity in front of them doubled… and closed. At Itaú Unibanco, Brazil’s largest bank, the same skill found two more: the bank’s venture arm and its innovation hub. Susheel booked both of those, too.

Six is a deliberate number: a person or two on each team that matters. Going wider costs more than it returns. “If I were to run the skill on twenty people, it would take an extremely long time and would dilute the quality of the emails,” he says. So CVC People Finder outputs a shortlist: the five people in the organization most likely to use Harmonic, weighted toward decision-makers like chief strategy officers, heads of innovation, and corporate development leads, each with their title, LinkedIn profile, seniority level, and a score. Susheel adds a sixth on purpose. “I’ve had success reaching out to very senior people, EVPs and C-suite,” he explains; “sometimes they can accelerate the demo booking or the sales cycle.”

“I’m perpetually surprised by the fact that Harmonic can find every single sub-sector of an organization,” Susheel adds. A corporation’s startup-facing units often have names that give no sign of the parent: Siemens’ venture arm is called Next47, and Susheel points to Deutsche Telekom’s incubator, hubraum. “It doesn’t have the word Deutsche or Telekom in it,” he says, “so you might otherwise miss it.” Harmonic resolves the relationship from its own map of corporate structures and live web research, so the arm comes back attached to the parent, no matter what it calls itself.

Prospects often assume a US platform thins out abroad. LVMH, Itaú, Deutsche Telekom and Airwallex are all bookings Susheel made outside the United States. 

‍

Writing an email worth replying to

Homing in on the company and the person is one thing; reaching out is another. Susheel’s second skill, CVC Email Generation, writes the email, beginning with research. “Firms will immediately discredit your email if you’re giving them companies that aren’t within their scope,” he explains. So the skill reads the firm’s website, its public content, and the articles written about it, and works out what kind of company that organization wants to meet. For a corporate venture arm, that’s an investment thesis. For an innovation unit or a corporate development team, it’s a focus area or an acquisition profile. The skill applies the same test to geography and stage, since a European industrial group has little use for a list of American seed companies.

Then the skill chooses which companies to put in front of that person. “Under-the-radar is important to me,” Susheel says. “These people obviously want to see companies they haven’t seen before, so outreach should offer them something of real value and preview what Harmonic can do.” As such, it weights toward companies recently out of stealth, without press, and with little public profile. Susheel chose the facts that get included alongside each company by asking Harmonic what its data has that other sources don’t. Those leading indicators are what’s available when a company is too young to have financials.

The email tells the recipient when Harmonic first indexed each company, often months or years before any other source. It breaks headcount growth out by department, which distinguishes a company still building its product from one that’s established a go-to-market motion. It surfaces an operator who left a well-regarded company, even if that operator hasn’t announced a next step. It notes whether a company is raising, on the founder’s own word, because founders submit that status to Harmonic directly. It names each company’s competitors and customers, and gives the recipient a first pass at burn or revenue. And it flags when notable investors and operators have started following a company’s team on X, which is investor interest surfacing before any announcement. 

“I gave the skill a format I wanted to follow exactly, and it follows it to the T,” Susheel says. Most AI writing tools drift, and one or two em dashes or a single generated-sounding phrase is enough for an email to read as slop. A draft that has to be cleaned of those tells costs more time than it saved. With Harmonic, “I don’t copy-and-paste the draft directly. I tweak it a bit, but the output is 99% of the way there.”

The signals that were, until recently, exclusive to venture investors are the ones Susheel’s recipients now need. A corporate development team choosing an acquisition target, or an innovation unit choosing which AI company to build with, has to decide while the field is still forming, and before a competitor locks the company down. 

‍

Selling with an investor’s tools

Susheel does the same work any growth analyst does against a named account list, but with the research behind each step specified once and running in minutes. He also uses Harmonic as a lightweight CRM, with contact details sitting on the same records as the lists, so nothing in the workflow requires a second tool. He sends more emails, to better-chosen people, with more meaningful data in each. Two months of working a target list produced eighty-three booked demos (more than two per working day), and more than a million dollars in pipeline before he ever carried a full quota.

None of the machinery that runs this was built for selling. Harmonic was built to answer two questions about any company anywhere: what’s happening there now, and who’s inside the company. That’s what an investor makes decisions on. Susheel pointed it at a sales territory using three skills he wrote himself in his first weeks on the team.