Segment-channel fit — when the buyer doesn’t live on LinkedIn.

The LinkedIn-presence test

Before the operator commits a LinkedIn account architecture, a proxy stack, or a multi-touch sequence to a target segment, the segment is run through a falsification test: does the buyer actually transact on this channel? The test has four operationally measurable inputs and a binary output.

• Profile-completeness rate. Pull 100 randomly sampled prospects from the segment and audit profile depth. A segment in which fewer than 60% of buyers have a complete profile, a recent post, and a connection count above 200 is not transacting on LinkedIn. SMB-owner ICPs typically land at 20-40% on this metric; corporate VP-of-Engineering ICPs land at 85-95%.
• 30-day login proxy.The “active in the last 30 days” signal in Sales Navigator and the recency of the most recent post or comment are the cleanest available proxies. Below 30% recency, cold InMail reply rates collapse from a 3-7% baseline to under 1% on the same copy.
• Acceptance-rate ceiling on a warmed account. Send 100 connection requests with a clean opener from a profile-complete account to the segment. An acceptance rate below 15% is the floor signal — the segment is either not on the platform or is over-saturated with cold outbound and has trained itself to ignore unknown senders. Mid-market software lands at 25-40%; SMB owners frequently land at 5-12%.
• Reply-rate-per-accepted-connection. Of the connections that accept, the rate that replies to a follow-up message within 14 days. Below 8% post-accept reply rate signals the segment treats LinkedIn as a credentialing platform rather than a communication channel, and the channel will not produce meeting volume regardless of copy.

A segment that fails three of the four tests is not a LinkedIn segment. The operator’s standard response — “the messaging must be wrong” — is the wrong diagnosis, and another four weeks of copy iteration on a channel the buyer does not transact on produces the same outcome.

The per-segment channel-presence map

The channel that produces pipeline is the channel the buyer already uses for transactional decisions. The map below is the empirical distribution by buyer segment, calibrated against the per-channel reply-rate floors at which the channel is operationally viable. A channel is “primary” if the segment produces a 2x or larger reply-rate-per-spend versus the next channel, “viable” if it produces meetings but is not the highest-yield channel, and “noise” if it produces a sub-1% reply rate on a warmed account.

The pattern that runs through the rows: LinkedIn is primary where the buyer’s career identity is built on LinkedIn, which is overwhelmingly the corporate-software buyer at director-and-above. Outside of that segment, LinkedIn falls to viable-at-best and frequently to noise. The default outbound stack is calibrated to the first two rows, which is why operators targeting any of the bottom seven find the standard playbook returns 0.3-0.8% reply rates and conclude the offer is wrong.

The SMB outbound stack

When the segment is SMB-owner, the stack that replaces LinkedIn is a four-channel architecture in which paid social does the top-of-funnel work LinkedIn does in the corporate stack, email and SMS handle the nurture, and phone — AI-dialed or human-dialed — handles the meeting-conversion step that calendar links handle in the corporate stack.

Paid social as top of funnel

Meta is the dominant SMB-owner discovery surface. The unit of work is a lead-form ad targeting the firmographic cohort, with an offer calibrated to the segment’s operational pain — a free template, a benchmark report, a calculator — that captures the lead’s phone number and email. The CPL on a well-targeted SMB-owner audience ranges from $8-25 for a low-friction lead-magnet offer to $40-120 for a meeting-request offer. The lead form bypasses the LinkedIn-versus-Workspace channel debate entirely because the prospect self-identifies into the pipeline rather than being identified by the operator.

The complement on the discovery surface is local-search and Google-services ads for segments where the buyer searches for a category (HVAC, plumbing, dental practice management). The CPL is higher ($40-150) but the intent signal is stronger because the buyer is in an active-search state at click time.

Email and SMS for nurture

Email functions as it does in the corporate stack — sequenced, authenticated, deliverability-aware — but the inbound source is the paid-social lead rather than a scraped prospect list. The reply rates on a paid-social-sourced email sequence run 4-8x the cold-outbound floor because the recipient has already raised their hand. The sequence is shorter (2-3 touches over 7 days rather than 5-7 over 28 days), the copy is offer-anchored rather than problem-naming, and the CTA is a phone-call slot rather than a calendar link.

SMS is the channel the corporate stack omits and the SMB stack relies on. The open rate on a TCPA-compliant SMS to a self-identified lead runs 90%+ within 4 hours; the reply rate runs 15-30% on a well-written first message. The operational constraints are non-trivial: a registered short code or 10DLC long code, an explicit double-opt-in capture at the lead-form step, a 1-call-to-action message structure, and a strict suppression-on-STOP discipline. The legal cost of a non-compliant SMS program is a TCPA class-action exposure of $500-1,500 per unconsented message, which produces a cost ceiling no other channel imposes.

Phone — AI-dialed and human-dialed

Phone is the channel the SMB buyer expects and the corporate stack treats as deprecated. The connect-rate floor on a warm SMB-owner lead is 25-45% within the first 4 hours of the lead-form submission, decaying to 8-15% by 24 hours and to under 5% by 72 hours. The speed-to-call window is the dominant variable in the conversion equation, which is the operational argument for AI-dialed first-touch on every paid-social lead inside the first 90 seconds.

The AI-dialed first call is a qualifier and a router, not a closer. The script identifies the buyer (versus a gatekeeper or wrong number), qualifies on 2-3 disqualifying criteria, and either books a meeting on the spot or routes to a human-dialed second touch. The conversion-to-meeting rate on AI-first then human-second is 12-25% on a self-identified lead, against 3-7% on a pure-email nurture and under 2% on a calendar-link-only follow-up.

The human-dialed touch is reserved for the qualifier-passed subset, which compresses the human cost from one full dial-and-disposition cycle per lead to one cycle per 3-5 leads. The math behind the speed-to-call window and the AI-as-router architecture is the central economic argument for spending the engineering hours on the lead-routing pipeline rather than on additional sequence variants.

Why founders default to LinkedIn anyway

The LinkedIn-first default is a sociological artifact, not an outcome of channel analysis. The founders writing the outbound playbook are themselves corporate-software buyers; their network, their content consumption, and their own buying behavior all happen on LinkedIn. The presumption that the buyer must therefore also live on LinkedIn is rarely tested before the account architecture, proxy stack, and sequence infrastructure are built.

The cost of this default on a misfit segment is high. The standard LinkedIn infrastructure — 8-12 secondary accounts, residential proxy stack, sequence tool, profile-warmup labor — costs $4-8K to stand up and $2-4K monthly to operate. On an SMB-owner segment that fails the LinkedIn-presence test, the infrastructure produces a single-digit count of meetings per quarter, against a paid-social-plus-phone stack at equivalent cost that produces 40-80 meetings per quarter on the same segment.

Cost-per-meeting across stacks, by segment

The per-stack cost-per-meeting is the cleanest decision-making artifact for the channel-mix question. The table below is the empirical range at a steady-state 90-day run-rate, after the warmup and learning costs have been amortized.

The ratio between the right-stack and the wrong-stack cost-per-meeting is typically 5-15x. No copy iteration, sender warmup, or sequence-cadence tuning closes a 5-15x gap. The operator who has been running the LinkedIn-first stack on an SMB segment for two quarters and has not produced meaningful pipeline is not looking at a messaging problem; they are looking at a stack-segment mismatch.

Where the channel decision sits in the planning sequence

The channel-mix decision sits between segmentation (Chapter 05) and intent-data integration (Chapter 07). It cannot precede segmentation because the channel-presence map is per-segment; a multi-segment list requires a per-segment channel decision rather than a single stack. It must precede intent-data integration because the intent signals that are operationally available (and the vendors that supply them) vary by channel — search intent feeds paid-social and email stacks; engagement intent feeds LinkedIn stacks; review-site and local-search intent feed phone-first stacks.

The channel-mix decision is also the gate on the infrastructure spend. The LinkedIn account architecture (Cluster 02), the email-domain procurement workflow (Cluster 04), and the paid-social ad operations spend are each five-figure commitments at the low end and six-figure commitments at scale. The operator who builds out all three before the channel decision pays for two-thirds of an infrastructure that produces nothing on the chosen segment.

The SMS regulatory footnote

SMS as a B2B channel is the part of the SMB stack with the highest legal exposure and the highest operational discipline cost. The TCPA regime, the 10DLC carrier registration, the per-campaign brand-vetting process, and the per-message content-filtering rules combine to produce a setup window of 6-10 weeks and a per-message cost of $0.0075-0.015 after carrier fees. The operator who treats SMS as a marketing add-on rather than a regulated communication channel produces either a non-deliverable campaign (carrier filtering) or a class-action exposure (consent violation). The compliance overhead is non-negotiable and is the reason most outbound-as-a-service vendors omit SMS from the default stack.

Common operator failures

• The LinkedIn-default failure.The operator runs the LinkedIn-presence test informally (“everyone is on LinkedIn”) rather than against the four-input falsification test, commits to the LinkedIn-first infrastructure, and produces single-digit quarterly meetings on a segment that requires a different stack.
• The channel-as-copy-problem misdiagnosis. The operator interprets a sub-1% reply rate as a copy problem, iterates through 4-6 variants over 8 weeks, and produces the same result because the buyer does not transact on the channel. The diagnostic that distinguishes a copy problem from a channel problem is the per-variant reply-rate variance: a copy problem produces a 3-5x spread across variants; a channel problem produces a flat sub-1% across all variants.
• The paid-social-without-phone failure. The operator runs paid social for SMB, captures leads, and routes them into a 5-touch email sequence on a 28-day cadence. The lead converts at 1-2% rather than the 12-25% the AI-dialed-plus-human stack produces. The speed-to-call window has decayed before the second email lands.
• The SMS-as-marketing failure. The operator treats SMS as an SMTP-class channel and either skips the 10DLC registration (producing carrier filtering and a sub-30% delivery rate) or skips the explicit opt-in (producing TCPA exposure on every send). The viable SMS program requires the same compliance discipline as the email-bulk-sender program in Cluster 01.
• The single-stack-for-multi-segment failure. The operator runs one stack across three segments to amortize the infrastructure cost. The two misfit segments produce 80% of the spend and 10% of the pipeline. The correct response is either to drop two segments or to run two stacks; running one stack on three segments produces the worst outcome on each.
• Treating phone as deprecated. The operator inherits the corporate-software prior that phone is for boomers and refuses to build the dial infrastructure. On any segment in which phone is the primary channel, the corporate stack converts at one-tenth of the phone-first stack on the same lead source.

Pre-deployment checklist

• The LinkedIn-presence test is run on a 100-prospect sample per segment, with the four inputs documented and the pass/fail recorded
• Each segment in the target list has an assigned primary channel and a viable second channel, justified by the per-segment channel-presence map
• The infrastructure spend is sequenced after the channel decision, not before — no LinkedIn-account procurement, no domain procurement, no Meta Business Manager build until the channel for that segment is committed
• For segments where phone is primary, the speed-to-call infrastructure is built before the lead-form ads run — AI-dialer integration, human-dialer rotation, disposition routing
• For segments where SMS is in the stack, the 10DLC registration, the double-opt-in capture, and the suppression-on-STOP discipline are operational before the first send
• The per-segment cost-per-meeting is measured at week 6 and week 12, against the per-stack benchmark band, with a written escalation rule for misses outside the band
• Multi-segment lists are run as multi-stack motions, not as a single-stack-amortized motion, unless two adjacent segments share the same primary channel

Where this fits

Segment-channel fit is the bridge from the upstream ICP work into the downstream infrastructure clusters. The segmentation chapter (Chapter 05) defines the cohorts; this chapter assigns each cohort a channel mix; the email-infrastructure cluster (Cluster 01), the LinkedIn cluster (Cluster 02), the conference cluster (Cluster 03), and the multi-channel orchestration chapter (Cluster 05, Chapter 09) are the downstream build-outs for the channels the assignment selects. The operator who treats the channel mix as a default rather than a decision pays the infrastructure cost of all channels and the conversion cost of the wrong ones.