# The Marketing Line — turn one operating memo into a governed workflow

Profile: flagship-master-16x9
Classification: public
CTA: Start the Relay Operator Workshop · $99 founding

## Transcript

### Chapter 1 — The glue tax

00:00:00.000–00:00:56.000

Every agency pays it — the glue tax. The quiet hours stitching leads to campaigns: a spreadsheet here, an export there, a dashboard that guesses. Until nobody can say which campaign actually produced this lead.

If you run an agency, attribution is the question you can never quite answer: which campaign actually produced this lead? In this lesson, you will build the Marketing Line in Relay — one bounded workflow that joins your CRM to your campaigns on a single shared field. You will inspect the join, shape the steps, place human judgment where it belongs, and keep evidence you can show a client. No black-box model. Two columns, one key, and a receipt.

Sources: source:website-story-glue-tax, source:marketing-line-memo, source:p3-presenter-script

### Chapter 2 — Inspect the memo

00:00:56.000–00:04:17.600

Here is where disciplined delivery starts — and it is not the automation canvas. It starts with an operating memo: one bounded process your agency already runs, written down. This is the Marketing Line memo. Before we open the product, we are going to read it properly.

Read a memo like a system map, not a story. Circle four things as you go: the records that exist, the fields that connect them, the state changes that matter, and the person who owns each decision. Four lenses — that is the whole method.

Start with the records. A lead lives in the C R M pack's leads table. A campaign lives in the Social pack's campaigns table. Two records, two owners — and in most stacks, two strangers. The lead came from somewhere, the campaign drove traffic somewhere, and nobody can prove the connection.

Now the fields. The useful detail is never the dashboard total — it is the shared value carried by two named fields. A campaign carries a U T M campaign value. A lead carries a source campaign value. Hold on to those two names; the entire next chapter stands on them.

Then the state changes, and the owners. A lead advances stage by stage. A campaign starts, runs, and completes. At every change that matters, the memo names the person who owns the decision — so when Relay runs this process later, human judgment already has a seat, on the record.

Now check the map against the product. These are real tables in Relay. Campaigns, owned by the social pack, with U T M campaign as a plain column. Leads, owned by the C R M pack, with source campaign. The memo's field names are the product's column names.

One discipline before anything gets built. The memo owns the operating idea. Relay owns the runnable steps. Keep that seam visible. When the client changes the idea, or the runtime changes the steps, you want the change on the surface where it happened — not buried inside a prompt.

So that is the read: two records, two named fields, owners on every decision that matters. What we have not done yet is prove the connection. Next, we isolate the join — the one sentence this whole system stands on — and watch it hold on live values.

Sources: source:marketing-line-memo

### Chapter 3 — Isolate the join

00:04:17.600–00:07:30.000

Here is the sentence this whole lesson stands on. A campaign carries a U T M campaign value. A lead carries a source campaign value. When they match, that lead is attributed to that campaign — and the match is a value you can read, sort, and back up.

Most tools answer the attribution question with a chart drawn by a heuristic you cannot inspect. When a client pushes back on the numbers, there is nothing to audit, because the join was never real. In Relay, the join is data. Two columns, one shared key. No attribution model between you and the truth.

Watch the join assemble. One card is the campaign, carrying U T M campaign. One card is the lead, carrying source campaign. The same string lands in both, the join line draws, and attribution becomes a fact you can point to.

Now look at it in the product, with live values. This is the Campaigns table the social pack owns. Every campaign row carries its U T M campaign value in a plain column — not a tracking pixel, not a report. A field you could sort right now.

And this is the Leads table the C R M pack owns. A lead that came in through a tagged link carries source campaign — the same string its campaign published. Two tables, two owners, one shared key. The handoff between them is testable.

So when a client asks why this lead sits in that campaign's funnel, you do not defend a model. You point at a value. Read it, sort it, back it up — and the evidence stays attached to the answer.

One honest edge. The join is only as good as the tagging upstream. Send a link out untagged, and the columns never match — the lead stays visibly unattributed. Relay will not invent a connection that was never real. For an agency, that refusal is exactly what keeps your numbers defensible.

Install either pack alone and these attribution features stay dormant. Merge them, and the bindings light up: two functions become one funnel, joined on a single key. Next, we shape that join into a workflow you can run, pause, and inspect.

Sources: source:marketing-line-memo

### Chapter 4 — Shape the workflow

00:07:30.000–00:10:56.600

The join is proven. Now shape the process that runs on it. In Relay, a workflow is not a prompt with steps hidden inside — it is a first-class record. This is the Marketing Line follow-up: the process your memo already named, written as stations anyone on your team can read.

Open it in the builder. Three stations on one line. Gather the attributed leads — the ones whose source campaign matches. Draft one follow-up per lead. Then send. Each station is one small job with its own prompt, and the output of one station becomes the context for the next.

Now the part that earns a client's trust. The send station carries one flag: requires approval. That is the checkpoint pattern — the moment the work is about to leave your workspace and touch a real inbox, the line pauses for a person. Judgment sits before the consequence, not after it.

Picture the line as a floor. Stations run in order, and before the send station stands a gate. Work stacks up behind it until a human lifts it. The gate is not a speed bump — it is the design: every consequential step waits for a decision, on the record.

Run it. The first two stations do their work and finish — leads gathered, drafts prepared. Then the line reaches the gate and stops. The status says waiting, and waiting means exactly that: the engine will not advance, and it will not send, until someone answers.

The question lands in the Inbox — one queue for every decision that needs a human, across every workflow you run. The checkpoint names the workflow and the station it is holding. Read what the agents prepared, then sign the gate: approve it, once, on the record.

Back on the workflow, the line has run through. Every station shows its result, and the send station shows something better: the follow-ups went out after operator approval. The record keeps your judgment exactly where a client would ask to see it.

That is the shape: stations you can read, and a gate where judgment belongs. What we have not done yet is treat the run itself as evidence. Next, we run the line and keep the receipt — same input, same receipt — and see what that discipline does to your margin.

Sources: source:marketing-line-memo, source:relay-workflow-guide, source:field-manual-workflows-chapter

### Chapter 5 — Keep the receipt

00:10:56.600–00:14:27.600

The gate lifted, and the line ran through. Most tools would call that done. Your client asks a harder question — did it work? In Relay you answer that before the run starts, by declaring what success means.

Open the workflow and read its success bar: three criteria, declared in advance. Run completed — required. Follow-ups sent after operator approval — required. Finished within two hours — advisory. Required misses fail the receipt; advisory misses mark it at risk. The bar is on the record before the work runs.

Now look at what the run wrote: one Operations Receipt, kept on the workflow itself. The verdict reads Passed — all three criteria met — and either way, the receipt names the next action. Not a dashboard to interpret. A verdict you can read in one line.

Open the evidence. Criterion by criterion, expected against observed: the run completed, the result says after operator approval, the duration beat its ceiling. And when evidence is missing, the receipt does not guess — it says at risk and names the missing fact. Nothing silently passes.

Proof is half the answer; the other half is cost. Cost and Usage is the ledger behind the work: metered spend for the month, a budget the engine cannot cross, and the headroom left before the cap — metered on your machine, not estimated on someone else's bill.

Every run lands in the audit log with its cost and its tokens. There is your follow-up line: three station steps, a few cents each, linked back to the workflow that ran them. The dollar has a paper trail.

And spend rolls up to the client, where margin lives — every run and token this client cost you, in one number beside the projects that spent it.

In the launch story, the founder shows one real client: six runs, thirty-two thousand tokens, five cents of model cost — inside a five-hundred-dollar monthly retainer. Five cents against five hundred dollars. That is the margin, in plain view.

One discipline holds all of this together: the receipt is deterministic. The criteria are snapshotted with the run, so editing a future success bar cannot rewrite what this run meant. Same input, same receipt — a record you can hand over, not a screenshot you have to trust.

That is the whole loop on one line of work: read the memo, prove the join, shape the process, gate the send — then run it, keep the receipt, and see the margin.

Sources: source:relay-operations-receipts-feature, source:relay-cost-usage-feature, source:website-story-receipt, source:website-story-margin

### Chapter 6 — Start the workshop

00:14:27.600–00:15:06.600

That is the whole discipline. The memo owns the operating idea. Relay owns the runnable steps. The join is a value you can read — and every run leaves a receipt you can show a client. No black-box model between you and the truth: one bounded workflow, two named fields, and evidence that survives the handoff. When you are ready to run it yourself, start the Relay Operator Workshop.

Sources: source:marketing-line-memo, source:website-offer-cta