Trigger orchestration and account-level attribution.
The chapters before this one produce a box in a courier van. This chapter is what separates a gifting program from a gifting expense: every physical event fires a digital action, so the channels reinforce each other instead of fighting — and the results are scored at the account, not the contact, because the person who opens the box is frequently not the person who books the demo.
TL;DR
- Hold shipment state as a CRM field —
sent → shipped → delivered— updated by fulfillment-vendor webhooks, never by a human checking a tracking page. - On SHIPPED, the account enters a paid-social audience targeted by title across the whole buying committee, not just the recipient. You don’t control who intercepts a package at an office.
- On DELIVERED, a same-day email fires from the human whose face is on the note card: “We sent a package to your desk at [address] — did it land?” Then the call-plus-email sequence starts.
- Build the recovery flow before launch. “I never got it” is the easiest conversation in sales — apologize, re-send to the corrected address, pitch while you’re at it.
- Instrument every package with a unique QR or link. Each scan confirms delivery, times follow-up, and feeds account intent scoring.
- Attribute at the account level. Contact-level attribution reports failure while the campaign works. Measure pipeline created at touched accounts against gift-plus-delivery spend.
The orchestration principle
The failure mode of most direct mail programs is not the gift, the list, or the copy. It is latency. A package delivers on Tuesday, the rep notices on Friday, the follow-up email lands the following Wednesday, and by then the gift is a paperweight with no memory attached to it. The window in which a physical delivery makes a digital touch feel warm is measured in hours, and no human-in-the-loop process reliably hits a window measured in hours across hundreds of accounts.
The fix is a single rule, applied without exception: every physical event triggers a digital action. Shipment turns on ads. Delivery fires the email and starts the sequence. A scan confirms receipt and re-times everything downstream. Nothing in the campaign waits on a human noticing that something happened in the physical world. Humans design the flow once, before launch; software runs it per-account, per-event, for the life of the campaign. This is the architecture Clay demonstrated publicly in its “How Clay Uses Clay” webinar series for its own ABM gifting motion, and it transfers to any team with a fulfillment vendor that exposes webhooks and a CRM that can hold a status field.
Orchestrated this way, the channels reinforce rather than compete: paid social warms the account while the box is in transit, the email lands the day the box does, and the call opens against a prospect who has encountered the company three ways in a week. Run un-orchestrated, the same three channels read as three unrelated vendors interrupting the same person.
The state machine
The mechanical core is a three-state machine held as a field on the account (or the recipient contact, mirrored up to the account) in the CRM — Salesforce, HubSpot, or equivalent. Fulfillment vendors with API access, &Open and Sendoso among them, emit webhook events as the package moves through the carrier network; each event advances the field, and each field transition is the condition that automation watches.
| State | Set by | Triggers |
|---|---|---|
sent | Fulfillment order placed via vendor API | Internal only: the account is locked against duplicate sends and enters the tracking queue. |
shipped | Vendor webhook on carrier pickup | The account enters the paid-social audience; buying-committee ads begin. |
delivered | Vendor webhook on carrier delivery confirmation | Same-day email from the card’s sender; multi-step call-and-email sequence enqueued; manual LinkedIn touches assigned. |
The field must be written by webhooks, not by a person reconciling tracking numbers. A human-updated status field is three days stale, and three days stale collapses the delivery-day email — the single highest-converting touch in the campaign — into an ordinary cold email that happens to mention a package.
On shipped: advertise to the buying committee, not the recipient
The moment the carrier scans the package, the account enters a paid-social audience — LinkedIn in the canonical build — targeted by title across the full buying committee at that account: CMO, CRO, RevOps, finance, whatever set of titles signs and influences a deal in the category. Not just the named recipient.
The rationale is that you do not control who intercepts a package at an office. Front desks sign for boxes; executive assistants open them; a curious teammate walks past while the box sits unopened. Whoever ends up holding the gift should already have seen the brand in their feed that week, so the physical object confirms something ambient rather than introducing something foreign. The goal state, in the recipient’s own words, is “I keep seeing this company everywhere” — the package lands as the third or fourth impression, not the first.
“Sniping” ads at only the recipient wastes precision the package already delivers. The package is the one-to-one channel; paid social is the one-to-committee channel, and spending it to re-target the one person already getting a box pays twice for the same impression while the other four members of the buying committee hear nothing. The per-account spend is small — the audience is title-filtered inside a single company — and it runs from shipment through the end of the outreach sequence.
On delivered: the same-day email and the sequence behind it
The delivery webhook fires the email the same day — automated, but written and addressed as one human to another. Two construction rules are non-negotiable.
First, the sender is the person whose name and face are on the note cardinside the box. If the card is signed by a founder, the email comes from the founder; if it carries an SDR’s face, the SDR sends. The recipient is about to hold a physical card with a name on it and read an email from a name — if the two differ, the campaign reads as an assembly line and the warmth evaporates. The account executive who owns the account is looped in as reply-to, so a response lands with the person who will run the deal.
Second, the body is short and asks a physical-world question: “We sent a package to your desk at [address] — did it land?” The address is included deliberately: it proves the sender did real work, gives the recipient something concrete to verify, and surfaces address errors immediately, feeding the recovery flow. The email is not a pitch. Its only job is to convert a delivery event into a reply.
Behind the email, the delivery event enqueues the standard multi-step sequence — calls and emails through whatever sequencer the team runs (Gong Engage or equivalent) — plus manual LinkedIn touches from the same sender. The continuity rule holds through the whole sequence: one face, one name, from note card to final touch.
The rep’s-eye view: manufacturing warmth
It is worth being precise about what the gift actually buys, because over-crediting it produces entitled follow-up that burns the account. The gift buys ten seconds and an opener. Nothing more. It does not obligate the prospect to take a meeting, and a rep who behaves as if it does converts a warm moment into resentment.
What the ten seconds change is the shape of the first live conversation. A cold call opens with a pitch the prospect must decide whether to tolerate. A post-delivery call opens with a question the prospect can answer warmly — “did the box make it to your desk?” — a question about their world, not the seller’s. Operators call this manufacturing warmth: the call begins at the temperature a second or third conversation would normally have. From there the rep runs entirely normal discovery — the gift got the conversation started; it does not carry it.
The recovery flow is a designed path, not an apology
Some fraction of packages will miss. The recipient works from home three days a week. The office moved. A coworker signed for the box and took it to another floor. An un-instrumented campaign books these as losses. An orchestrated one books them as its easiest openers, because “I never got it” is the easiest conversation in sales: the prospect has initiated contact, about a gift, with a mild grievance the seller can resolve in one step.
The flow: apologize briefly, confirm the correct address in the same exchange, trigger a re-send through the fulfillment vendor — and, since the prospect is engaged and talking, optionally advance the conversation while you’re at it. A prospect who has just told you where they actually sit has done it in a thread they started.
The operational requirement is that this path exists before launch: a one-click (or fully automated) re-send against the vendor API, budgeted in the campaign economics, with the corrected address written back to the CRM. A recovery flow improvised per-incident takes a week per incident, and a week is long enough for the warmth to expire. Designed in advance, a failed delivery is an opener. Improvised, it is a loss with extra steps.
The unique-link spine
Every package carries a unique QR code or personalized link — per package, not per campaign. This is the tracking spine of the entire program, and it does three jobs at once.
It confirms delivery independently of the carrier — carrier webhooks confirm a box reached a building; a scan confirms a human opened it and engaged. The gap between the two is exactly the population the recovery flow exists for. It times the follow-up — a scan is the strongest possible send-now signal, and sequences that re-time against scan events consistently outperform delivery-date offsets. And it feeds account intent scoring— every scan and page visit is an intent event written against the account, including visits from sessions that are clearly not the recipient. Which is where this chapter’s second pillar begins.
Account-level attribution
The second pillar of this chapter is a measurement rule: attribute at the account level, or the data will tell you the campaign failed while it works.
The mechanism is simple. Gifts are social objects inside an office. The recipient may never reply to a single touch and still cause the deal: they hand the gift to a teammate, mention the company in a meeting, or forward the delivery-day email to whoever owns the problem. The person who books the demo is then a contact the campaign never touched, and a contact-level model assigns the meeting to “inbound” while assigning the gift a zero.
A concrete example from Clay’s own campaign: one target executive never responded to any touch. His package’s unique link, however, logged six visits — and in the following weeks, teammates from his company began registering for webinars. At the contact level, that account is a dead row. At the account level, it is the campaign working precisely as designed: the gift entered the building, circulated, and generated engagement from the buying committee it was always aimed at.
The scoring model that follows: every touched account is flagged for an attribution window (a quarter is typical). Within it, count intent events (scans, link visits, webinar registrations, inbound from any contact) and pipeline events (meetings, opportunities, closed-won) at the account, regardless of which contact produced them. The recipient is the delivery mechanism; the account is the unit of analysis.
Measuring: pipeline per dollar, benchmarked against cold
The campaign-level metric is pipeline created at touched accounts divided by campaign spend, where spend is the gift cost plus delivery and fulfillment fees. Ad spend and rep time can be tracked separately, but the core ratio deliberately uses the incremental physical cost, because the comparison that matters is against the channel the same accounts would otherwise get: cold outbound.
Cold outbound has a pipeline-per-dollar figure too — most teams have simply never computed it. Compute both and the gifting decision stops being a taste question: a $20 gift that materially multiplies meeting-book rates on tier-one accounts pays for itself in rep efficiency alone; the same gift sprayed at unqualified accounts loses to cold email on every line. The instrumentation in this chapter exists in large part so this ratio is computable at all. An un-instrumented campaign produces anecdotes; an instrumented one produces a number the CFO can compare against every other dollar of GTM spend.
The amplification layer: the campaign about the campaign
The last orchestration layer costs almost nothing and routinely out-reaches the packages themselves: publicly teaching the motion you just ran. Executive posts walking through the campaign design, a webinar on how it was built, the creative itself shared as an artifact — each of these reaches an audience orders of magnitude larger than the shipment list, and some of that audience are the very accounts being targeted, now encountering the company a fourth way.
Clay’s numbers here are instructive: three executive LinkedIn posts about the campaign generated over 200,000 impressions, and the motion earned an unsolicited pickup in a major marketing newsletter — reach no per-package budget could buy, from a campaign that shipped a few hundred boxes. The amplification layer also compounds the “everywhere” effect: a buying-committee member who saw the ad, then the box, then the founder’s post explaining the box, has constructed the impression of a company with far more surface area than it has spend.
The only rule: amplify the craft, not the targets. Posts teach the mechanism; they never name recipient accounts or publish anyone’s engagement.
A worked timeline
| Phase | Work |
|---|---|
| Weeks 1–2 | List finalized and scored; office addresses researched; recipients selected; remote workers filtered. All triggers, audiences, templates, sequences, and the recovery flow built and tested against staging webhooks. |
| Week 3 | Assets ship: gifts ordered through the vendor, note cards printed with the sender’s name and face, unique links generated per account, paid-social audiences created (empty, waiting on shipment events). |
| Week 4+ | Triggers run. Webhooks advance states; ads, emails, and sequences fire per-account as events arrive. Human work shifts entirely to conversations and the daily queues below. |
From week 4, the daily operational rhythm is three queues, each reviewable in minutes:
- The scan digest — accounts that scanned or visited in the last 24 hours, so reps prioritize live-right-now accounts for calls and LinkedIn touches.
- The delivery queue — packages confirmed delivered, with the same-day email verified sent and the sequence verified enqueued. The check is that automation fired, not to fire it manually.
- The recovery queue — deliveries stalled past the expected window plus “never got it” replies, each awaiting an address correction and a one-click re-send.
Common operator failures
- Human-triggered follow-up. The rep watches tracking and sends “when it lands.” It lands Tuesday; the email goes out the following Monday. The delivery-day touch is forfeited on every account.
- Sniping ads at the recipient only. The one person guaranteed a one-to-one touch gets the ad budget too, while the rest of the buying committee never hears the name.
- Sender discontinuity. The card carries the founder’s face; the follow-up arrives from an SDR alias. The one-to-one illusion collapses and the sequence reads as automation.
- No recovery path at launch. “I never got it” replies queue for days while someone figures out how to re-send. The easiest conversation in sales expires unanswered.
- Campaign-level links. One shared QR across all packages confirms nothing, times nothing, and scores nothing. The spine is per-package or it does not exist.
- Contact-level attribution. The recipient never replied, so the account is marked dead — while teammates visit the landing page and book through the website. The campaign is cancelled for “not working” at the moment it is working.
- Skipping the amplification layer. The team runs a genuinely clever motion and tells no one, leaving the cheapest reach in the program unclaimed.
Pre-launch checklist
- CRM status field created;
sent → shipped → deliveredtransitions written exclusively by fulfillment webhooks, verified end-to-end against test orders - Paid-social audience defined by buying-committee titles per account; entry wired to the
shippedevent; exit wired to sequence completion - Delivery-day email templated with address merge field; sender identity matched to the note-card face; account executive set as reply-to
- Call-and-email sequence and manual LinkedIn tasks enqueued off the
deliveredevent in the sequencer - Recovery flow live: address-correction capture, one-click re-send against the vendor API, corrected address written back to the CRM, re-send budget reserved
- Unique link or QR generated per package, mapped to account, with scan events writing to account intent scoring
- Attribution window and account-level scoring model agreed with sales leadership before the first box ships
- Cold-outbound pipeline-per-dollar baseline computed, so the campaign has a comparison number waiting
- Amplification assets planned: at least one executive post and one teach-the-motion artifact scheduled for the in-flight window
Where this fits
This chapter consumes everything the cluster built before it: the account selection and office-address research that decided where boxes go, the creative that decided what lands, and the fulfillment vendor whose webhooks make the state machine possible. Vendor selection (Chapter 04) is the hard dependency — a vendor without event webhooks caps this entire chapter at manual reconciliation, which is reason enough to disqualify one.
It is also where the program justifies itself. Gifting without orchestration is a cost center with charming anecdotes. Gifting with a state machine, a buying-committee audience, a designed recovery path, and account-level scoring is a channel with a pipeline-per-dollar figure that can stand next to cold outbound in a budget review — and, run well against the right accounts, beat it.
Related chapters
- Fulfillment and vendors — the vendor API and webhook capabilities this chapter’s triggers depend on.
- Gifting and direct mail — all chapters — the cluster hub.
- The ABM direct mail playbook — the end-to-end play this chapter’s orchestration layer plugs into.
Allston Labs operates the full sending estate as a service.
We provision domains, configure the entire authentication record set, run warmup, and monitor reputation across providers. The stack lives under your entity. The engineer on call lives in your Slack.