How we run outbound for a product nobody has searched for yet

Social Intel is prelaunch. Nobody types "social intel" into Google, and nobody searches for a category name we invented for a product they have never used. Inbound will arrive later, after the content compounds. The first fifty conversations have to come from somewhere else.

So we run outbound. This post covers how, because the mechanics are more useful to another founder than another post about how AI is changing everything.

Disclosure before anything else. The outbound engine belongs to Ripe Leads, a done-for-you B2B lead generation agency in Vilnius. Ripe Leads and Social Intel share a founder. That makes this a working note from inside a group of companies, not a review by an unrelated customer. Read it that way, and weigh the claims accordingly.

Why we did not build the engine ourselves

The tempting version is obvious. Buy a few domains, rent a sending tool, scrape a list, write a sequence. Every step looks like a Saturday.

The steps that do not look like a Saturday are the ones that decide whether any of it lands. Domain warm-up runs for weeks before a single real send. Verification has to catch the catch-all addresses that quietly wreck a sender reputation two weeks later. Someone has to watch placement, rotate mailboxes, keep the DNS records honest, and notice the day replies stop arriving because a provider started filtering. That work never finishes and it never shows up in the product.

We are a small team building a product with a scraping pipeline, a transcription queue and an analysis layer. Every hour spent on mailbox hygiene is an hour not spent on the thing people are being asked to buy. Renting the engine was the cheaper trade.

The signal we target

Most outbound targeting starts with a firmographic filter. Company size, industry, country, job title. That gets you a list of companies who could plausibly buy, which is a different thing from a list of companies with a reason to answer this month.

The filter we care about is behavioural. Social Intel reads what a market is publishing and reports which hooks, formats and topics are earning attention. A team only feels that pain if two things are true at once:

• They already publish on a schedule, so a bad month costs them something visible.
• They already research competitors by hand, so the work exists and someone resents it.

An agency running content for six clients hits both. A brand entering a new language market hits both, usually with a deadline attached. A company that posts twice a quarter hits neither, and no subject line fixes that.

Finding companies that match a behavioural filter is harder than filtering a database. You have to look at what the company actually does in public. That is the same research problem the product solves, which is a pleasant symmetry and also the reason we understood the brief.

What the agency handles

The split runs roughly like this. Ripe Leads owns targeting, data, deliverability, copy and sending across cold email and LinkedIn, and forwards interested replies. We own the reply itself, the call, and everything after it.

That last part matters more than founders expect. An agency can produce a conversation. It cannot produce your product knowledge, your pricing judgement, or your willingness to tell someone the tool is wrong for them. Outsourcing the top of the funnel while ignoring the middle produces a stack of warm replies that go cold in your inbox.

We also feed the copy. The agency writes it, and we supply the specifics that make it worth reading: what a scan actually returns, what it costs, what it will not do. Generic copy about "unlocking growth" gets ignored in every market and every language.

The languages question

Our first markets are the Baltics and the German-speaking region. Ripe Leads runs campaigns in Lithuanian, Latvian, Estonian, Polish, Czech, Slovak, German, English and Russian, which is the practical reason the arrangement works at all.

Translated English reads as translated English. A German marketing lead can tell within one line whether a message was written for them or run through a translator, and the second kind gets deleted. If you sell into Europe and your outbound is English-only, you are competing for attention in every prospect's second language while your local competitors write in their first.

What we refuse to publish

You will not find a reply rate on this page. Or a meeting count, or a pipeline figure, or a screenshot of an inbox with the names blurred out.

One prelaunch campaign, in two markets, for a product that has not opened to the public, produces a number with no predictive value for anyone else. Publishing it would make this post more persuasive and less true. Ripe Leads takes the same position in their own sales conversations and refuses to promise a fixed number of meetings, which is the part of their pitch that convinced the founder who now writes their cheques.

When there is a result worth reporting, with enough volume and enough time behind it to mean something, it will appear here with the method attached so you can argue with it.

What we would do differently

Two things, in hindsight.

We defined the ICP too broadly in the first pass. "Marketing teams and agencies" is not a target, it is a category. The behavioural filter above came later, after a round of replies from companies who found the idea interesting and had no reason to act. Interesting is not a buying signal.

We also underestimated how much product detail the copy needed. The early drafts sold the outcome. The versions that earned replies described the mechanism, because the buyer we want has already been sold an outcome by four other tools this quarter.

If you are running the same play, the next two posts are the ones I would read: building an ICP from public evidence, and what to do when the list is right and the message is wrong.