The 48-hour campaign · our end-to-end production timeline
Hour by hour, this is how we take a campaign from a one-line brief to live across email, paid, and social in 48 hours.
Why 48 hours is the number
Most agency campaigns take two to three weeks. The brief sits in a queue, a strategist writes a doc, a designer waits on the doc, copy waits on design, and the whole thing limps to a kickoff call before anyone touches a real asset. We run Arthea differently. We have 83 autonomous agents wired into one operating system, and the slow parts of that chain are now machine work that happens in parallel. The human job is judgment. The assembly is handled.
We settled on 48 hours because that is the point where speed stops being a constraint and starts being a moat. A client can react to a competitor, a news beat, or a sales dip on the same week it happens. The campaign is live while the situation is still live. Below is the actual timeline we run, with the hours marked and the handoffs called out. Nothing here is a best case we hit once. This is the standard cadence, and the reason we can promise it is that the bottlenecks that used to define campaign timelines have moved off the critical path.
A quick note on what 48 hours covers. It is brief to live across email, paid, and at least one social channel, with a landing page and full tracking. It does not cover net-new brand identity work or a video shoot, because those genuinely need more time. For the kind of campaign most accounts run weekly, a promotion, a launch, a seasonal push, a reactivation, 48 hours is the real number.
Hours 0 to 4: brief, research, and the offer
The clock starts when a brief lands. The brief is short on purpose. One paragraph: who we are talking to, what we want them to do, and the constraint that matters most (budget, deadline, or a specific product push).
In the first four hours, three things run at once. A research agent pulls the account's last 90 days of performance data from the CRM and from Klaviyo, so we know which segments are warm and which subject lines already worked. A second agent scrapes the competitive set and the live landing pages, so the angle we pick is not one a competitor is already running. A human strategist reads both outputs and writes the offer in plain language, because the offer is the one thing a model should not invent for you. By hour four we have a single page: audience, offer, channels, and the one metric we will judge this on.
That offer page is the spine of everything downstream. Every copy variant, every ad, every email references it. We learned the hard way that if the offer is fuzzy at hour four, the campaign is fuzzy at hour 48, and no amount of fast production fixes a vague promise. So we spend real human time here even though it is the slowest single step. Getting the offer right is the highest-leverage hour in the whole timeline.
Hours 4 to 16: copy, design, and the build
This is the block that used to eat two weeks. We split it into channels and run them in parallel.
Copy agents draft the email sequence, the paid variants, and the social posts from the same offer page, so the message is consistent before a human edits a word. We typically generate 4 subject-line variants per email and 6 ad-copy variants per platform, then a copy lead cuts that to the 2 we will actually test. Design runs against a locked component library, so a landing page is assembled from existing blocks rather than drawn from scratch. A typical campaign landing page is live in staging by hour 12.
The build matters here. We do not hand a Figma file to a developer and wait. The page goes straight into the client's Webflow or our own stack, the email goes into Klaviyo as a draft flow, and the paid assets go into the ad manager as paused campaigns. By hour 16 everything exists in its real home, unpublished. This is the single biggest difference from a traditional process. There is no separate handoff step where a static design has to be rebuilt as a working page. The thing we review is the thing that will ship, in the environment it will ship from.
The component library is what makes this safe to do fast. Because the page is assembled from blocks we have already tested for responsiveness and accessibility, the agent is composing known-good parts rather than inventing layout. That removes an entire class of bugs before the human ever looks at it. The creative variation lives where it belongs, in the copy, the imagery, and the offer, instead of in re-solving how a hero section should stack on mobile.
Hours 16 to 36: review, QA, and the human gate
We sleep on it, deliberately. The campaign sits overnight and a fresh set of eyes reviews it in the morning. This is the qualitative gate, and it is always human. We check that the offer reads as true, that the design holds at 375px and 1280px, that every link resolves, and that the tracking fires.
The QA pass is partly automated. A test agent walks the landing page, confirms the pixel and click tracking, sends a seed email to a deliverability inbox, and checks that the paid campaigns target the segments the brief named. Anything it flags goes back to the relevant agent with the specific line to fix, and the fix lands in minutes. By hour 36 we have a campaign that a human has approved end to end.
Hours 36 to 48: launch and the first read
Launch is a sequence rather than a single button press. We publish the landing page, activate the email flow, and move the paid campaigns from paused to live in a set order so that traffic always has somewhere to land. We hold a small budget cap on paid for the first few hours so we can read early signal before we spend the full daily.
Then we watch. The first read comes at the six-hour mark after launch: open rate on the email, click-through on paid, and bounce on the landing page. If the landing page is bouncing above what the account normally does, we know the offer-to-page match is off and we patch the page rather than the ad. By hour 48 the campaign is live, instrumented, and already producing the data that shapes the next version.
A real example
A skincare client came to us on a Tuesday morning with a slow week and a warehouse full of a product they needed to move. The brief was one line: clear the surplus to the segment most likely to repurchase. By Tuesday afternoon we had the offer, a bundle priced to protect margin. By Wednesday midday the landing page, a 3-email flow, and paid social were built and in staging. We reviewed Wednesday evening, launched Thursday morning, and had the first sales read by Thursday afternoon. The campaign cleared the surplus inside the week instead of sitting on a roadmap.
The part that mattered was not the build speed. It was that the offer was right. The research agent had surfaced that the warm repurchase segment responded to bundles rather than single-unit discounts, and that one finding shaped the whole campaign. A slower process would have produced a prettier asset and a worse result, because the prettier asset would still have been built on the wrong offer. Speed only compounds when the early judgment is sound, which is exactly why we keep humans on the offer and the gate and let agents have everything in between.
What we measure
We judge a 48-hour campaign on one primary metric chosen at hour four, never a vanity number. For a reactivation it is revenue from the dormant segment. For a launch it is qualified add-to-carts. For a clearance it is units moved against the surplus. The whole point of naming the metric before we build is that nobody gets to redefine success after the numbers come in. The campaign either hit the thing we said it would or it did not, and the next version is built on that answer.
What makes the speed real
The timeline is not magic and it is not a sprint where everyone skips sleep. It works because the boring half of campaign production is now agent work running in parallel against locked assets, and the human half is concentrated on the two decisions that actually move outcomes: the offer and the qualitative gate before launch. We are an AI systems and growth engineering team, and this is the kind of compression that shows up when the system sets the pace instead of the calendar.
If you want to see what a 48-hour campaign looks like for your account, that is the work we do every week at arthea.ai.
