An agency's problem is not finding leads for one client, it is running the same discipline for eight at once without the accounts bleeding into each other. That means a repeatable intake, a source budget per client, qualification that runs without you, and a monthly report a client can read in ninety seconds.
Disclosure: Signl is ours, and agencies are a large share of who uses it. Limits and behaviours named here were checked against our code and published policies on 24 October 2026. The process below works with other tooling, or with a spreadsheet and more hours.
Key facts
- The scarce resource in client prospecting is attention per account, not volume of leads.
- Three to six sources per client is the working range. More produces noise you will not read.
- A source budget is the number to manage. Signl's plan is 25 sources, and a search counts as 2.
- Separate suppression per client is not optional: one client's objector must not reappear through another.
- Reporting renews retainers. Quoted comments from named buyers beat every chart.
Intake: the six answers you need before adding a source
Most agency outbound fails at intake, not at execution. Each of these is a question a client can answer in a call, and each one maps to a setting.
- What do you sell, in one sentence a stranger would understand? This becomes the ICP's offer statement, and it is the thing qualification is judged against.
- Which titles and seniorities buy it? Not "decision makers". The actual job titles on the last ten deals.
- Which countries can you invoice? A hard filter. Signl caps the match when a known country falls outside the list, whatever else looks good.
- Company size, floor and ceiling. The ceiling matters more than agencies expect; enterprise logos in a mid-market pipeline distort everything.
- Who are your three closest competitors? Needed for sourcing and for reading dissatisfaction.
- Whose posts do your buyers read? Clients often do not know. Start with their own and one competitor, and find the rest from data after a fortnight.
Signl can propose an ICP from the client's website and you then correct it, which is faster than a blank form and roughly as accurate as the website is. The field reference is in define your ICP.
Ten minutes on the ICP saves a month of bad leads. This is the single highest-leverage part of agency delivery and the part most commonly rushed into a shared template.
The source budget
Treat sources as a budget, because they are both the cost driver and the noise driver.
| Source | Counts as | Who it suits |
|---|---|---|
| Client's own profile | 1 | Every client. Warmest people, and the easiest opener |
| Client's team's posts | 2 (it is a search) | Clients with 5+ people posting |
| One or two competitors | 1 each | Every client with a named competitor |
| A creator the buyers follow | 1 | Clients in a category with real commentators |
| One topic search | 2 | Clients whose buyers describe the problem in stable language |
| Mentions of a competitor | 2 | Competitive categories, switching-heavy markets |
A typical client lands at 4 to 8 source units. On Signl's 25, that is three to five clients on one plan before you need a custom arrangement, which past 25 sources is agreed by hand.
Which brings up the thing to settle before you build a business on it: our fair use asks customers not to share one account across multiple organisations or resell access, so if you are running client work, talk to us first and we will structure it. That is not a sales line, it is the clause, and we would rather say it on this page than in an email three months in. The same question exists with every vendor in this space; ask it of all of them.
Qualification you can defend to a client
The agency version of qualification has an extra requirement: when a client complains about a lead, you need to be able to show why it passed.
So keep three things per person, and make sure your tooling does too.
- The comment, verbatim. It is the evidence, and it is what the opener answers.
- A reason in words. Not a score alone. "Head of finance at a 200-person SaaS company, said month-end close takes nine days" is defensible. "Confidence 72" is not.
- The rejections, with reasons. This is where you find out your client's ICP was wrong, before they do.
Signl runs discover, classify, enrich, compose in that fixed order, which matters commercially for an agency: a person the ICP check filters out never consumes a verified email lookup, so a client's allowance goes only to people who fit. The plan includes 500 verified work emails a month, with no per-person charge; why paying for attempts rather than results is bad maths is in the bounce tax.
Delivery: three service shapes
Agencies run this one of three ways, and the choice determines your margin and your risk.
| Shape | What you deliver | Where the risk sits |
|---|---|---|
| Qualified lists | People with context, into the client's CRM | Lowest risk, lowest price, easiest to churn out of |
| Lists plus drafted first lines | The same, with a grounded opener each | The sweet spot for most agencies. Risk is tone |
| Full outreach | You send, from the client's domain or account | Highest price, and you own their deliverability and their reputation |
If you send, send from the client's tooling, not yours. Signl prepares, qualifies and routes people and the connected tool does the sending, which keeps the account and the domain reputation where it belongs. Nine destinations exist today: Instantly, Smartlead, lemlist, HeyReach, Expandi, HubSpot, Clay, a generic webhook and Slack alerts. Native LinkedIn messaging from the client's own account is listed as coming soon on our pricing and is not something to sell as live.
Keep autopilot conservative on client accounts. Rules can route people at or above a confidence from chosen source types into a campaign with a daily cap, up to 10 rules and 500 sends a day per account. Our advice is to run a client's first month with the rules off, read the openers yourself, and turn on one rule for the client's own-post source only.
Separation: the five things that must not leak
- ICP. Per client, always. A shared ICP produces leads that fit nobody.
- Suppression and objections. One client's objector must never surface through another client's source. In Signl an objection suppresses and erases the person globally and permanently, for every customer, which is stricter than per-client and is the behaviour you want.
- Source labels. The own, competitor, creator and topic relationship governs what an opener may say. A competitor source labelled as own is how a client's prospect learns they were watched.
- Credentials. The client's sequencer keys are the client's. Signl stores pasted credentials server side and never repeats them back.
- The leads themselves, at the end. Agree in writing who keeps them when the retainer ends.
Reporting that renews retainers
One page, monthly, in this order.
Per source, over 30 days: people collected, people who fit, people contacted, replies. Four columns, one row per source. The ratio between the first two columns is the conversation to have: a source producing 400 people and 3 fits should be dropped, and the client should see you drop it.
The three best conversations. Name, company, and the comment they actually wrote. This is the part clients read.
What you changed. One source dropped, one added, two ICP tweaks. Visible process is what distinguishes a retainer from a subscription.
What you did not get. The honest line. If a source produced nothing, say so in the report rather than waiting to be asked.
Do not report engagement volume, impressions or follower growth. They are not your job and they invite the client to judge you on something you do not control.
Pricing the service, and the obvious competitor
Your own cost base here is small: a tool, plus the hours. The hours are what you are selling, and the published market is wide. Cleverly's guide, published 11 August 2025 and modified 3 June 2026, puts done-for-you organic outreach from $397 a month and agency retainers at $3,000 to $25,000 a month, with pay-per-appointment at $50 to $500 per qualified meeting (read 24 October 2026). SalesBread publishes $3,000 a month plus a one-time setup fee for roughly 20 or more qualified leads a month (read 24 October 2026).
We sell a managed service ourselves, at $2,000 a month, so take this as interested: the agencies that hold retainers longest are the ones whose reports show process, not the ones promising a lead count. A guaranteed number of meetings is a promise about someone else's buying behaviour.
If you want to see how we structure the managed version, it is on done for you, and a buyer-side view of the whole build-versus-buy decision is in done-for-you LinkedIn lead generation.
Limits
- Only public engagement is available. No follower lists, no connections, no post viewers, from any vendor.
- Automated collection is contractually contested. LinkedIn's User Agreement section 8.2 prohibits scraping and bot access (read 24 October 2026). If you run this for clients you are making that decision on their behalf, which is worth putting in the contract. The separate data protection question is in GDPR and B2B prospecting. This is not legal advice.
- Client-side bottleneck. The usual failure is not lead quality, it is a client who does not reply to their own replies. Agree who handles responses at intake.
- This does not scale by adding sources. It scales by adding clients with tight ICPs. Twelve sources on one client produces a list nobody reads.
