Sofia Marchetti is a senior technical recruiter at a 250-person robotics startup, where she is the only person sourcing for eight open engineering roles. The candidates she wants are employed, busy, and bombarded with generic outreach. A single channel no longer reaches them. The email she sends sits unread, the LinkedIn message gets lost in a crowded inbox, and a cold text without permission is both annoying and, in the United States, potentially illegal. Sofia's job is not to pick the best channel. It is to orchestrate email, LinkedIn, mobile messaging, and video into a sequence that meets each candidate where they actually pay attention, while staying inside the rules that govern each one.

Why One Channel Is No Longer Enough

Sofia used to be an email purist. She wrote careful, personalized notes and waited. The problem was simple math: a strong cold email to passive engineers might earn a single-digit response rate, which means the large majority of her best prospects never replied at all, not because they were uninterested but because they never saw it or never got around to it. Adding channels is not about being noisier. It is about giving the same well-crafted message more than one chance to land, on the surface where a particular candidate happens to be reachable. Each channel has a personality, a set of norms, and a legal framework, and using one as if it were another is the fastest way to look careless.

That distinction between noise and surface area is the whole discipline, and it is easy to lose. A recruiter who reads "multichannel" as permission to contact the same person four ways in two days has not added surfaces; they have added volume, and the candidate experiences it as one sender being difficult to escape. Multichannel done properly means each contact arrives at a different moment, in a different register, with something new attached, on a surface the candidate might plausibly be checking. The candidate should experience a sequence of reasonable approaches from someone who is persistent. They should never experience being surrounded.

Email: The Workhorse, Governed by CAN-SPAM

Email is Sofia's workhorse for detailed, asynchronous outreach. It is the right channel when she needs to convey real substance about a role, the team, and the mission, and when she wants the candidate to be able to reply on their own schedule. Her best practice is a short, specific subject line, a first sentence that proves she read the candidate's actual background rather than scraped their job title, and a single clear ask. She keeps the body under roughly 150 words because a wall of text on a phone screen gets archived.

The single clear ask does more work than it appears to. An email offering three possible next steps, a call or a coffee or a look at the job description, hands the candidate a decision before they have decided they are interested, and the easiest resolution to a decision you did not ask for is to close the tab. One ask means the reply is a yes or a no rather than a small project. The roughly-150-word ceiling serves the same end: a message that fits on a phone screen can be read in the short gap a busy engineer has between meetings.

Email also carries a legal floor. In the United States, the CAN-SPAM Act governs commercial email, and recruiting outreach that promotes opportunities can fall under it. The practical obligations are accurate header and subject information, a clear way to opt out, honoring opt-out requests promptly, and including a valid physical postal address. Sofia builds these into her templates once so that every send is compliant by default rather than something she has to remember candidate by candidate.

The template placement is the point worth copying. Compliance that depends on remembering fails on exactly the day the recruiter is busiest, which is also the day the most messages go out. Put the postal address and the opt-out into the template signature block and they appear on every message including the hurried ones, at no ongoing cost. The accurate-header-and-subject requirement constrains the writing rather than the footer: a subject line engineered to look like a reply to a conversation that never happened is the behavior the rule exists to prevent, and it is a bad idea on its own merits since the candidate finds out within one sentence.

LinkedIn: Where Passive Candidates Live

LinkedIn is where Sofia's passive candidates already are, in professional mode, which makes it ideal for a warm, low-pressure first touch. She uses LinkedIn Recruiter to send an InMail when she has no email address, and she favors a connection request with a personal note over a hard InMail pitch when the candidate is more than one degree removed. The norm here is brevity and relevance: a two or three sentence message that names a specific reason she reached out to this person, not a paragraph that could have gone to anyone.

The preference for a connection request over an InMail at distance is a judgment about what each format asks of the recipient. A connection request with a personal note asks for a small, reversible yes and opens a channel; an unsolicited pitch from a stranger several degrees away asks the recipient to evaluate an opportunity from someone they have no reason to trust yet. Sofia is not avoiding InMail because it performs badly. She is sequencing the ask so the first thing she requests is the cheapest thing to grant, which is also why she keeps the note to two or three sentences: a long note attached to a connection request has already broken the format's implicit promise that this will be quick.

The strategic value of LinkedIn is context. Sofia can reference a project, a talk, or a shared connection, and that specificity is what separates a reply from a dismissal. She treats LinkedIn as the opening of a relationship rather than the close of a deal, and she moves the conversation to email or a call once the candidate signals interest, because LinkedIn messaging is a poor place to negotiate details.

The tell for when to move is the candidate's first substantive question. Up to that point the exchange is short and well suited to the platform. Once they ask what the compensation band looks like or how the team is structured, the answer needs paragraphs, may need forwarding to a partner, and will need to be findable in three weeks, none of which a messaging thread supports. Sofia answers briefly and offers to send the detail by email, which respects the question and relocates the relationship to a surface that can hold it.

Mobile and SMS: High Response, Strict Consent Under TCPA

Text messaging has the highest and fastest response rate of any channel Sofia uses, and it is also the one with the sharpest legal teeth. In the United States, the Telephone Consumer Protection Act, or TCPA, governs text messages, and the core rule is consent. Sofia does not cold-text candidates. She obtains prior express consent first, typically when a candidate has applied and agreed to text communication, or when they have explicitly opted in, and she always offers a clear way to stop, such as replying STOP. A cold recruiting text to someone who never opted in is not just bad manners, it can create real liability for her employer.

It is worth being precise about what consent means here, because the word gets used loosely. Having someone's mobile number is not consent. Finding the number on a resume is not consent, since the candidate supplied it for the application rather than for recruiting texts on unrelated roles. Consent is the candidate agreeing to be contacted this way, recorded so that Sofia can show when and how it was given. That recording requirement is the practical part: a consent nobody logged is a consent nobody can demonstrate later, which for a rule with real liability attached is functionally the same as not having it.

Because consent is the gate, Sofia treats SMS as a mid-sequence and late-sequence channel rather than a first touch. Once a candidate is engaged and has agreed to texting, she uses it for exactly the things text is good at: confirming an interview time, sending a quick reminder, or a short nudge that respects the candidate's time. She keeps texts brief, identifies herself and her company every time, and never uses the channel for anything sensitive or detailed.

The restriction to logistics is a design choice rather than a limitation. Text arrives with an urgency the medium cannot modulate: it interrupts, it is read within minutes, and it offers no room to soften or contextualize. That makes it excellent for "your interview moved to 2pm" and poor for anything a candidate might need to sit with. Sensitive news by text reads as someone avoiding the harder conversation, and detail by text is detail the candidate cannot find again. Self-identifying on every message answers the medium's other weakness, which is that a text from an unknown number carries no context at all.

Video: The Differentiator for Personal Touch

Video is Sofia's differentiator. A 30 to 60 second personalized video, where she says the candidate's name and references their specific work, stands out precisely because almost no one sends them. She uses video sparingly and deliberately, reserving it for high-priority candidates or for a re-engagement attempt when text-based channels have gone quiet. The norm is authenticity over polish: a genuine, slightly imperfect video reads as human, while an over-produced one reads as a marketing asset. Video also shines later in the process, for example a short personal note congratulating a candidate or walking them through next steps, where the warmth of a face and voice carries more than a paragraph ever could.

The authenticity norm is worth defending, because the instinct on seeing a rough take is to record it again. What a personalized video signals is that a specific human spent a few minutes on this specific person, and polish undercuts exactly that signal: a smooth, evenly-paced video is indistinguishable from one recorded once and sent to everyone on a list. The small imperfections are the proof of work. The one thing that must be right is the specific reference, because a video that says the candidate's name and nothing else is a form letter with a face attached.

Sparingly is the other operative word. Video works because it is rare, so a recruiter who sends one to every candidate has spent the differentiator and is left with a slower way of sending an email. The 30-to-60-second ceiling protects the same scarcity from the other side: a video long enough to require a decision about whether to watch it will not be watched.

Sequencing and Cadence: Orchestrating the Channels

The skill is not any single channel, it is the sequence. Sofia designs a multi-touch cadence that spaces contact out, varies the channel, and stops the moment a candidate asks her to. A typical sequence for a passive engineer runs over about two weeks: a LinkedIn connection request with a personal note on day one, a detailed CAN-SPAM-compliant email on day three if there is no LinkedIn reply, a brief value-add follow-up email on day seven, and a short personalized video on day ten as a final, differentiated touch. SMS enters the picture only after the candidate has engaged and consented, where it confirms times and keeps momentum.

Read the last column as the load-bearing one. The sequence is not four channels in an arbitrary order; it is an escalation from cheapest ask to most effortful touch, gated where the law requires a gate. Day one asks for a small reversible yes on the surface the candidate already uses professionally. Day three spends real substance on those who did not take it. Day ten spends Sofia's own recorded minutes on a shortlist. And SMS, the channel with the highest response rate, sits outside the acquisition sequence entirely, because what makes it powerful, arriving directly on the candidate's phone, is exactly what makes it require consent first.

Sofia personalizes the substance, not just the merge field. AI tools help her draft and adapt the same core message for each channel's length and tone, but she edits every one so the specific hook about the candidate survives. She also respects cadence as a courtesy: too many touches in too few days reads as desperation, while a well-spaced sequence reads as persistence. And the most important rule across every channel is that a no, or a request to stop, ends the sequence immediately.

The adaptation step is where multichannel campaigns most often lose their quality, because compressing a 150-word email into three sentences for a connection request is a summarization task, and summarization drops specifics first. What survives compression by default is the generic frame, the company, the role, the enthusiasm, while the one sentence about the candidate's actual work is the easiest thing to cut. Sofia's edit inverts that: the hook is the part that must survive at every length, and the frame is what gets cut to make room. A three-sentence message that is all frame is the version nobody replies to.

The stop rule is absolute and cross-channel. A candidate who declines on LinkedIn has declined, and emailing them on day three because the decline arrived on a different surface is the same disregard the rule exists to prevent, dressed as a technicality. In practice the stop has to be recorded somewhere that governs the whole sequence rather than living in the thread where it happened.

A Worked Example: One Sequence Across Four Channels

Sofia runs a sourcing campaign for a senior controls engineer role, reaching out to 100 matched passive candidates with her four-channel sequence. The illustrative results show why orchestration beats any single channel. The day-one LinkedIn connection note with personalization earns replies from about 22 of the 100. The day-three email to the roughly 78 who did not respond on LinkedIn pulls in another 9 replies, a response rate near 12 percent on that step. The day-seven follow-up email adds a handful more. The day-ten personalized video, sent to a shortlist of the 20 highest-priority non-responders, lands 5 replies, a 25 percent rate on that small, carefully chosen group. SMS, used only with the candidates who engaged and consented, drives near-immediate confirmations for interview scheduling and almost no drop-off.

Follow the arithmetic across the steps, because the shape of it is the lesson. The three steps with counted replies contribute 22, then 9, then 5, which is 36 candidates in conversation, and the day-seven follow-up's handful is what carries the sequence to the roughly 40 of 100 the campaign reached in total. No individual step performed spectacularly. The largest single contribution came from the opening touch and still left more than three quarters of the list untouched by it, and every step after that was working a smaller and smaller pool. The total is not the product of one channel working well; it is the product of four modest results not overlapping.

The video step is the one that rewards a closer look, because its two numbers point in opposite directions and both are true. Five replies is the smallest count in the campaign, and a 25 percent rate on the step is the highest per-message result Sofia gets. That is not a contradiction; it is what targeting looks like. The video went to 20 candidates she had already identified as highest-priority, on day ten, after three prior touches had established some familiarity with her name, and every one of those conditions is doing work. Reading the 25 percent as a property of video and concluding that she should send video first, to everyone, would destroy all three conditions at once and produce a worse result than the email step it replaced.

No single channel would have reached even the LinkedIn total alone. The sequence as a whole moved roughly 40 of 100 passive candidates into a conversation, and the video step, despite its tiny audience, produced the highest per-message response rate because it went to the right people at the right moment. These numbers are illustrative, but the lesson is durable: match the channel to the stage, respect the consent rules that govern each one, and let the sequence, not any one message, do the work.

Anti-Patterns

Treating multichannel as permission to contact everyone everywhere. This is sending the LinkedIn note, the email, and the text within the same two days. It happens because more surfaces sound like more chances and a compressed sequence gets to the answer faster. What goes wrong is that the candidate does not experience four channels, they experience one sender who is hard to escape, and what was meant to read as persistence reads as desperation. The counter is Sofia's spacing, roughly two weeks with a different channel and a new reason to engage at each step.

Using one channel's norms on another. This is the four-paragraph pitch sent as a connection request, or the two-sentence note sent as a cold email. It happens because pasting one core message everywhere is faster than adapting it, and the adaptation feels cosmetic. What goes wrong is that the message signals the sender does not work on that surface: the long connection request breaks the format's promise that this will be quick, and the two-line cold email carries no substance to justify itself. The counter is writing to each norm, under roughly 150 words with one clear ask for email, two or three sentences for LinkedIn, brief and self-identifying for text.

Cold-texting because you have the number. This is treating a mobile number found on a resume as permission to text about an unrelated role. It happens because SMS has the highest and fastest response rate of any channel, which makes it the most tempting shortcut in the sequence. What goes wrong is that under the TCPA the gate is prior express consent, not possession of the number, and a cold recruiting text to someone who never opted in can create real liability for the employer. The counter is Sofia's placement of SMS mid and late sequence only, after the candidate has agreed to text communication, with the consent recorded and a clear way to stop such as replying STOP.

Leaving CAN-SPAM to memory. This is remembering the opt-out and the postal address on the messages you are thinking about. It happens because the requirements feel like footer trivia next to the work of writing a good message. What goes wrong is that the sends most likely to miss them are the hurried ones during the busiest week, when the most messages go out, and the accurate-header requirement gets violated by exactly the subject-line trick that seems clever under pressure. The counter is building accurate headers, the opt-out you honor promptly, and the valid physical postal address into the template once.

Polishing the video until it stops working. This is re-recording until the 30 to 60 seconds are smooth, well-lit, and evenly paced. It happens because the instinct on watching a rough take is that it is not good enough to send. What goes wrong is that polish is the signal of mass production: a flawless video is indistinguishable from one recorded once for a whole list, so removing the imperfections removes the evidence that a specific human spent time on this specific person. The counter is authenticity over polish plus one non-negotiable, the specific reference to the candidate's own work.

Letting a decline on one channel not stop the others. This is emailing on day three someone who declined on LinkedIn on day two. It happens because the decline lives in the thread where it arrived while the sequence is scheduled somewhere else, so nothing connects them. What goes wrong is that the candidate reads it as being ignored rather than as a systems gap, and their reading is the one that matters. The counter is a stop that governs the whole sequence rather than the channel it arrived on, recorded centrally and applied the moment a no lands anywhere.

Practice Prompts

• Audit your last twenty outreach messages by channel. Count how many went out on each surface and how many were adapted for it rather than pasted across. One core message on three channels at the same length is one channel in three places.
• Rewrite one email to Sofia's constraints. Cut a message you sent this month under roughly 150 words, with a specific subject line, a first sentence proving you read the candidate's background, and exactly one ask. Note what you had to remove.
• Compress that same email into a connection request. Get it to two or three sentences and check what survived. If the candidate-specific hook was the casualty and the company frame survived, invert the edit.
• Write down where your text consent comes from. For every candidate you texted last month, name when and how they agreed and where it is recorded. Any candidate you cannot answer for is the gap.
• Check your email template against CAN-SPAM. Confirm accurate header and subject information, a clear opt-out you honor promptly, and a valid physical postal address, and confirm they live in the template rather than in your habits.
• Record one 30-to-60-second video and do not re-record it. Send it to a high-priority candidate who has gone quiet, naming their specific work in the first ten seconds.
• Map your own two-week sequence. Write the channel, day, job, and consent requirement for each step, mark where SMS becomes available, then write the rule for what happens the moment a candidate says no on any surface.

Reflection

• Which channel do you default to, and is that because it works or because it is comfortable?
• If a candidate declined on one surface today, what in your process would stop the next scheduled touch on another?
• Where does your text consent come from, and could you show when and how it was given?
• Read your last connection request. Would you have accepted it from a stranger?
• What in your outreach survives compression from email length to three sentences, and is it the part about the candidate?
• When did you last send something to a candidate that asked for nothing at all?

Glossary

• Multichannel sequence. A planned series of touches across different surfaces, spaced over time, giving one well-crafted message several chances to land rather than adding volume.
• Cadence. The spacing of touches, treated as a courtesy the candidate can feel; too many in too few days reads as desperation, well spaced reads as persistence.
• Channel norms. What a competent professional looks like on a given surface, including length and register. Violating them signals the sender does not work there.
• CAN-SPAM. The United States act governing commercial email, which recruiting outreach promoting opportunities can fall under. Requires accurate header and subject information, a clear way to opt out, prompt honoring of opt-outs, and a valid physical postal address.
• One clear ask. Requesting a single next step, so the reply is a yes or a no rather than a decision the candidate has not agreed to make.
• Connection request with a personal note. The low-cost opening ask on LinkedIn, preferred over a hard InMail pitch at more than one degree because it requests a small, reversible yes.
• InMail. The platform message type Sofia uses when she has no email address for a candidate.
• TCPA. The Telephone Consumer Protection Act, which governs text messages in the United States and makes consent the gate for recruiting texts.
• Prior express consent. The candidate agreeing to be contacted by text, recorded so it can be demonstrated later. Possessing a mobile number is not consent.
• Stop keyword. The clear way out offered on every text, such as replying STOP, available regardless of how consent was obtained.
• Personalized video. A 30-to-60-second recording naming the candidate and their specific work, kept authentic rather than polished because polish is the signal of mass production.
• Value-add touch. A message carrying something useful without a new ask, which distinguishes a sequence from repetition of the same request.
• Cross-channel stop rule. A no or a request to stop, recorded centrally and applied to the whole sequence rather than only to the surface it arrived on.
• Per-message response rate. The reply rate on a single step, distinct from cumulative reach; the video step's high rate reflects a small, already-warmed audience rather than a property of the channel.

Related Lessons

• Hands-On Project: Design a Communication Campaign with AI Support builds a full sequence end to end, including the A/B testing and measurement this lesson's orchestration assumes.
• AI-Assisted Outreach: Templates, Personalization, and Quality is the drafting layer behind adapting one core message to each channel's length and tone without losing the hook.
• Personalization at Scale: When AI Enables Better Communication develops the difference between personalizing the substance and personalizing the merge field.
• Passive Candidate Engagement: Drafting with AI covers why the passive candidate rarely replies to a first message, which is the premise the whole sequence rests on.
• Avoiding Generic or Manipulative Messaging is directly relevant to the accurate-header requirement, since a subject line engineered to look like something it is not fails on both counts.
• Consent and Transparency: What Candidates Need to Know develops the consent standard behind the SMS gate and the record that makes it demonstrable.

Closing

Sofia's constraint has not changed: one recruiter, eight open engineering roles, and a candidate pool that is employed, busy, and already receiving more outreach than it reads. What changed is that she stopped asking which channel works and started asking what each channel is for. Email carries substance and a legal floor she builds into the template. LinkedIn opens the relationship at the lowest possible cost to the candidate. Text moves logistics quickly for people who agreed to be texted. Video spends her own minutes on the few candidates worth spending them on.

The sequence is what turns those four judgments into reach. Roughly 40 of 100 passive candidates entered a conversation not because any step performed spectacularly, but because four modest results did not overlap, arriving over two weeks in different registers with a new reason each time. That is the durable form of the lesson, and it holds regardless of which surfaces are fashionable when you read it: match the channel to the stage of the relationship, write to the norm of the surface you are on, honor the consent rule that governs it, and stop everywhere the moment a candidate asks you to.

Key Takeaways

• Orchestrate channels, do not just pick one. A single channel gives a well-crafted message one chance to land. Email, LinkedIn, mobile, and video together give it multiple surfaces, which is why a sequenced campaign reaches far more people than any one channel alone.
• More surfaces, not more volume. Multichannel means different moments, different registers, and something new attached each time. Contacting the same person four ways in two days is one sender who is hard to escape, not four chances to land.
• Email is the workhorse and must respect CAN-SPAM. Use it for substance, keep it under roughly 150 words with one clear ask, and build the legal requirements into the template: accurate headers, a clear opt-out you honor promptly, and a valid physical postal address.
• LinkedIn is for the warm first touch. Its strength is context. Name a specific, genuine reason in two or three sentences, prefer a connection request over a hard pitch at distance, and move to email or a call at the candidate's first substantive question.
• SMS has the highest response rate and the strictest consent rule. Under the TCPA, texting requires prior express consent, recorded so it can be shown. Possessing a number is not consent. Treat SMS as a mid-sequence logistics channel, always offer a clear way to stop, and never cold-text.
• Video is the differentiator, used sparingly and left unpolished. A 30-to-60-second video works because it is rare and because its imperfections prove a human spent time on this person. Reserve it for high-priority candidates and re-engagement, and always name their specific work.
• Sequencing and cadence do the real work. Space touches over roughly two weeks, vary the channel, and personalize the substance rather than the merge field, protecting the candidate-specific hook when you compress a message for a shorter format.
• The stop rule is absolute and cross-channel. A no on any surface ends the whole sequence immediately, which means the stop has to be recorded centrally rather than left in the thread where it arrived.

Frequently Asked Questions

Is four touches over two weeks too many? The number matters less than what each touch carries. Sofia's sequence spaces contact across roughly two weeks, changes channel at every step, and gives the candidate a new reason to engage each time rather than repeating the same ask in different words. Four repetitions of one request inside a few days is the version that reads as desperation, and it is a different thing wearing the same shape. Cadence is a courtesy the candidate can feel, and the one rule that overrides the schedule entirely is that a no, or a request to stop, ends the sequence immediately.

The video step had the best response rate. Should we lead with video? No, and the reason is visible in the conditions attached to that number. The 25 percent rate came from 20 candidates Sofia had already identified as highest-priority, contacted on day ten, after three prior touches had made her name familiar. Moving video to day one would remove the targeting, the warming, and the scarcity that produced the result, and would spend her recorded minutes on 100 people instead of 20. The high per-message rate is a property of that position in the sequence, not of the medium, which is exactly why the sequence and not the channel is the unit of design.

We have candidates' mobile numbers from their applications. Can we text them about new roles? Not on the strength of the number alone. Under the TCPA the gate is prior express consent to be contacted this way, which typically comes from a candidate who applied and agreed to text communication or who explicitly opted in, and a number supplied for one application is not agreement to be texted about unrelated roles. Sofia's practice is to obtain consent first, record when and how it was given so it can be demonstrated later, and always offer a clear way to stop such as replying STOP. A cold recruiting text to someone who never opted in can create real liability for the employer, not just irritation.

How do I adapt one message across channels without it becoming generic? Decide in advance what must survive, because compression drops specifics first. The natural casualty when a 150-word email becomes a three-sentence connection request is the one line about the candidate's own work, while the company-and-role frame survives because it is easier to summarize. Sofia inverts that: the hook is protected at every length and the frame gets cut to make room. AI tools help draft and adapt for each channel, but she edits every version herself for exactly this reason.