The ideal first customer is an operator agency that already pays for reply handling, in cash or in hours, and has not found a tool that does it across clients. The directory has about twenty of them: the 4s and 5s that run 10 or more clients on Smartlead or Instantly, with a documented reply problem or a homemade fix. The wedge and Go-to-market plan set the outline: 5–50 clients, 50k–500k sends a month. The 108 entries in the directory narrow it down.
Qualify on built glue and reply volume, not on complaints. A Make scenario, an Airtable reply log or a paid AI reply agent predicts willingness to pay better than a quote in a vendor case study.
The ICP
| Attribute | Target | Evidence from the list |
|---|---|---|
| Team | 2–20 people, founder still selling | Most bands are 2–10 (GTM engineers and Clay shops); RevenueBase puts 37 of 43 French cold-email agencies at 1–10 |
| Clients | 10–50 live clients, at least 3 on the desk in the pilot | Danish Lead Co. 110 served; RevGrowth 40+; StackOptimise 50+ (self-reported, cumulative) |
| Volume | 50k–500k sends a month | Danish Lead Co. 500k+; StackOptimise "2m+" is the upper edge |
| Stack | Smartlead or Instantly, ideally both; Clay upstream; API or webhooks in use | 69 of 108 on Smartlead, 37 on Instantly, 21 on two or more sequencers |
| Geography | Based in the UK, France, Ireland, the Nordics, the Baltics, Portugal or Benelux. Clients prospect into opt-out markets and the US | 26 entries are in countries where Julian may email them by name (EU/EEA country matrix) |
| Pain | Replies scattered across client workspaces; clients cannot see their own replies; per-client fees | 13 of 44 pain signals are about reply handling (The first 100 buyers) |
| Current workaround | A master-inbox routine, Slack alerts, Airtable or Make glue, a paid AI reply agent | GenFlows, Leadmoves, LeadHaste, UpscaleB2B, Growth Engine X |
| Budget | Spends roughly $1,500–2,500 a month on platform and inboxes for 10 clients; charges clients $1,100–5,000+ each | Hub arithmetic: ~$2,040/mo for 10 clients on Smartlead; client prices in European outbound agencies |
| Buying trigger | Adding clients past a plan cap, a sequencer switch, a client asking for reply access, the Microsoft SMTP AUTH deadline, hiring a setter | See below |
The desk's founding price is €149 a month, locked for 24 months (Pricing strategy), for the first 20 partners, against a list price of €199. By my estimate, one setter-day saved each month covers it. LeadHaste charges its clients $997 a month for an AI reply agent on top of a retainer. That is the price of reply handling on the agency's side of the bill.
Buying triggers worth watching
- The next client. Smartlead's $29 per client from Pro (as of October 2026) lands on every new signing (Lead-gen agencies).
- A second sequencer. Litehouse (Instantly + EmailBison) and Spring Drive (Smartlead + Instantly) already have two inboxes to watch.
- A client who wants to see replies. Experiment5m's clients "struggled to view and respond to campaign replies".
- A hire. Scalantec is hiring to serve 23 clients. A desk is cheaper than the next setter.
- 31 December 2026. Microsoft's Basic SMTP AUTH goes default-off for existing tenants. Agencies with Outlook inboxes will be shopping (Microsoft rules: Outlook.com and Microsoft 365 limits).
Anti-ICP
| Who | Why not now |
|---|---|
| German agencies emailing German prospects | §7 UWG makes that actionable from one email. The rule engine would block their core traffic (Germany) |
| Agencies that want the desk as compliance cover | Several sell into consent countries (Clientium, Vertriebsfrei, Reachflow). If the engine blocks their sends, they churn; if it does not, Julian carries the risk (Platform liability: what the sequencer itself risks) |
| Course graduates with 1–3 clients | Instantly's early churn was about 10% a month from users "not experiencing success" (The info economy) |
| Shops over ~20 staff or with in-house tooling | Long security reviews; Leadbird builds tools itself, and Inbox Accelerator and ProspectUp sell infrastructure |
| Consultancies where outbound is one module | mgsh., Peakora, Partner UP: the sender is a small decision for them |
| Explicitly satisfied users | Cymate: "I do not want to switch to another software" (Visible operators) |
| Creators with student audiences | Lead Gen Jay's audience is the high-churn cohort. A channel at best, later (Distribution) |
Discovery questions
Use these on a 30-minute call. The first five mirror the script in The first 90 days, so the answers feed Kill criteria K1 and K2.
- "What are the three biggest operational problems in running your clients?" Ask unprompted, then have them rank per-client fees, reply handling, client reporting, deliverability and support.
- "How many clients are live today, on how many workspaces and sequencers, and how many mailboxes?"
- "Walk me through yesterday's replies. Who saw them first, how long until a positive one got an answer, and where did out-of-office replies go?"
- "What have you built to cope: Slack alerts, Make, n8n, Airtable, an AI agent? What does it cost to maintain?"
- "How do you prove a meeting to a client? Has a client ever disputed one?"
- "Can your clients see their own replies today? Do they ask to?"
- "What do you pay each month for the sequencer, inboxes and reply tooling?"
- "Into which countries do your clients send? Has a client, DPO or lawyer asked whether it is allowed?"
- "What is your split of Google, Microsoft and private SMTP mailboxes? Do you know what changes for Microsoft on 31 December?"
- "If one queue held every reply from every client, with opt-outs suppressed everywhere, what would you stop doing?"
- "Would you put three client workspaces on it for 30 days, then pay €149 a month if it works?"
Disqualify on the call if: fewer than five live clients; no use of Smartlead, Instantly or EmailBison webhooks or API; core traffic into consent countries with no consent records; or the buyer's answer to question 10 is "nothing".
Fit-scoring rubric
The directory's 1–5 fit was set by three passes from public evidence. This rubric is written to reproduce those bands, so that scores from discovery calls can be compared with the published ones. Score each criterion 0, 1 or 2.
| Criterion | 0 | 1 | 2 |
|---|---|---|---|
| A. Multi-client load | Not an agency, or <5 clients | 5–20 clients | 20+ clients, or 2+ sequencers |
| B. Stack match | No named sequencer | One of Smartlead, Instantly, EmailBison | Two, or API/webhooks in documented use |
| C. Reply pain or glue | None public | Indirect (capacity limits, per-result billing, white-label portal) | Documented reply pain or a homemade reply layer |
| D. Shape | >20 staff, consultancy, or part competitor | Outbound is one of several services | Cold-email operator, 2–20 people |
| E. Signal quality | Site dead, or counter-signal | Directory listing only | Case study, own site and a named founder agree |
| Points | Fit | Meaning | Examples from the data |
|---|---|---|---|
| 9–10 | 5 | Design-partner candidate now | LeadLead BangBang, Danish Lead Co., Growth Engine X, Leadmoves |
| 7–8 | 4 | Strong fit with one gap | A-Sales, Litehouse, Halfwarm, Spring Drive |
| 5–6 | 3 | Plausible; needs a call | Clientium, BuildingReach, Peakora |
| 3–4 | 2 | Thin evidence or small wallet | Traction, Kadima Labs, Sellent |
| 0–2 | 1 | Counter-signal | Cymate (operators pass) |
Two adjustments come after the total. Reachability is scored separately, not added: a consent-country 5 stays a 5 but is routed to Where they gather. A call overrides public evidence. If questions 3 and 4 show no reply load, drop the score by two.
The passes scored by judgement, not with this table. Spot checks reproduce the bands for the examples shown, but a few entries will move by one point under the rubric. Cymate was scored 5 by one pass and 1 by another.
What this means for an entrant
- Write the ICP into the CRM. Log rubric scores A–E for every call so that K1 can be checked against fit, not anecdote.
- Look for glue before pain. A Make scenario or a $997 reply agent is a budget line Julian can take over.
- Keep half the partners at 10+ clients. Below that, the per-client argument is weak (Lead-gen agencies).
- Say no to compliance-cover buyers. The rule engine's credibility is worth more than one consent-country agency (EU-native compliant outbound).
- Re-score after every call and push the new fit back to the directory.