How to Use AI to Plan a Marketing Campaign in 2026: Step-by-Step Guide with Prompts

Learning how to use AI to plan a marketing campaign comes down to running six tool assisted steps in order: audience research with Perplexity, strategy with ChatGPT, copy with Jasper and Anyword, creative with Gamma and Canva AI, channels with Buffer, Mailchimp AI, and Zapier, and measurement with HubSpot AI. This guide walks through every step with copy paste prompts, exact prices from our 281 tool database, and the workflow habits that make the output feel planned rather than generated.

• The whole plan fits in a two week part-time sprint. Six sessions of one to three hours each produce buyer research, a documented strategy, a full copy set, campaign creative, scheduled channels, and a measurement plan, which is the scope agencies quote thousands of dollars to deliver.
• The planning discipline is the real advantage. CoSchedule surveyed more than 3,000 marketers and found that those who document their strategy are 414 percent more likely to report success, yet almost 40 percent still operate with no documented strategy, and this workflow produces the documented plan as a byproduct.
• The coordination problem is where AI earns its keep. The HubSpot 2026 State of Marketing Report found 52 percent of teams now run five to eight channels and 94 percent diversified their channels last year, and the AI stack in this guide keeps every asset quoting the same messaging pillars across all of them.
• The payoff numbers justify the effort. McKinsey research puts the revenue lift from personalization at 5 to 15 percent with acquisition costs cut by up to 50 percent, and Litmus data shows email returning $36 to $40 for every dollar spent.
• Top rated picks for this task: ChatGPT and Zapier lead the stack at 4.7/5 each, Buffer, Claude, and Gamma hold 4.6/5, and every tool in this guide offers a free tier, so the required investment before launch is your time rather than your budget.

Campaigns fail for boring reasons, and the planning data is blunt about it. CoSchedule research across more than 3,000 marketers found that organized marketers are 674 percent more likely to report success than their peers, that marketers who document their strategy are 414 percent more likely to succeed, and that proactive planners are 331 percent more likely to succeed, yet almost 40 percent of marketers still report having no documented strategy at all. A structured campaign plan attacks this gap directly, and AI collapses the cost of producing one: the strategy document, the messaging guide, and the channel schedule that used to take a planning week now fall out of the workflow as finished artifacts.

The second pressure is channel sprawl. The HubSpot 2026 State of Marketing Report found that only 6 percent of marketers rely on just one or two channels, 52 percent run five to eight, and 94 percent of teams diversified their channel mix in the last year, with 45 percent of them naming the move from broad platforms to niche communities as their most effective change. Every additional channel multiplies the assets, the variants, and the handoffs, which is exactly the kind of volume work AI absorbs: one positioning pillar becomes a LinkedIn post, an Instagram hook, an email intro, and a paid social headline without five separate writing sessions.

The third pressure is money. Gartner 2026 CMO Spend Survey reported marketing budgets effectively flat at 7.8 percent of company revenue, while CMOs now allocate an average of 15.3 percent of those budgets to AI initiatives, and HubSpot found 73 percent of marketers say their budgets are more heavily scrutinized than they used to be. There is a gap in that data worth naming: Gartner also found only 30 percent of organizations feel ready to scale their AI capabilities, which means the budget has arrived before the workflow literacy has. This guide is the literacy part, step by step, with the prompts included.

The payoff side of the ledger is just as concrete. CoSchedule State of AI research found 85 percent of marketers now use AI for content creation, that 83 percent report increased productivity since adopting it, and that AI saves marketers more than five hours every week on average. McKinsey research puts the revenue lift from personalization at 5 to 15 percent with customer acquisition costs reduced by as much as 50 percent, and email returns $36 to $40 for every dollar spent according to Litmus, the highest consistent ROI of any major channel. The rest of this guide shows you exactly how to capture that with six steps and a copy paste prompt for each one.

Before running the workflow, it helps to know the destination. A complete marketing campaign plan in 2026, whether the campaign runs for two weeks or a full quarter, contains eight standard components, and each one maps to a specific step in this guide so nothing is improvised the morning the campaign goes live.

• Audience and market research: who buys, what language they use, where they gather, and what alternatives they already compare you against. Handled in Step 1.
• Campaign strategy: the one measurable goal, the campaign concept, the ranked audience segments, and the channel shortlist with a budget split. Handled in Step 2.
• Messaging pillars: the two claims every asset repeats in the same words, tested against a skeptical customer before anything ships. Handled in Step 2.
• The copy set: the landing page, the social posts, and the announcement email, all written from the same pillars. Handled in Step 3.
• Creative assets: the campaign briefing deck and the platform sized graphics that make the campaign feel real. Handled in Step 4.
• Channel setup and automation: the scheduled posts, the email flow, and the Zaps that move leads between tools without manual work. Handled in Step 5.
• Measurement plan: the three to five numbers that define success, declared before launch rather than argued after it. Handled in Steps 2 and 6.
• Retrospective: the structured review that turns one campaign into the playbook for the next one. Handled in Step 6.

One practical note on execution: this workflow works solo, but it also shines for a two or three person team, because each step produces an artifact the next person can pick up without a meeting. The researcher hands the audience profile to whoever runs strategy, strategy hands the pillars to the copywriter, and the copywriter hands finished text to whoever owns scheduling and automation. The campaign hub in Notion AI makes the handoffs visible, which is why teams that adopt this sequence report fewer launch morning surprises than teams coordinating over chat threads.

If you skip a component, campaign week finds it. The most commonly skipped pieces are the measurement plan and the retrospective, and they are also the two that marketers most often say they wished they had. The workflow below takes roughly two weeks part-time, and every step produces a finished artifact before the next one starts, so the plan is documented before the first ad spends a dollar.

This step produces the evidence layer that every later step quotes, so doing it first prevents the most expensive campaign mistake, which is building assets around assumptions. First, open Perplexity, switch the mode to Pro Search, and enter the following prompt, replacing the example product with your own:

I am planning a three week marketing campaign for a meal prep subscription in Austin, Texas. Build the audience picture: the three problems my target customers have with dinner planning, the exact phrases they use to describe those problems, the social platforms and communities where they spend time, and the five closest competing options with their prices and positioning claims. For every fact, give me the source name, the year, and a direct link, and flag anything where sources disagree.

Run a second prompt focused on the channel landscape, because the campaign concept and the channel shortlist come from it:

For a meal prep subscription launching a campaign in Austin, list which local channels respond best to new food brands: neighborhood groups, local newsletters, food bloggers, gym communities, or office catering programs. Note any seasonal food trends in the next three months and two local events the campaign could piggyback on. Cite a source for every claim.

Save both answers into one research document with the links intact, because Step 2 and Step 3 will quote them directly. When you have real customers to interview, record the calls and let tl;dv transcribe and summarize them on its free plan, which offers unlimited recordings across Zoom, Google Meet, and Teams with timestamped highlights you can quote in copy later. For teams that want search demand data on top of the qualitative picture, Semrush Copilot analyzes the competitive content landscape and topic gaps, though at $139.95 per month for the Pro plan it is the priciest tool in this guide, and honestly more than a first campaign needs, so treat it as the upgrade for round two rather than the entry point.

Perplexity is free to start with Pro at $20 per month, and tl;dv offers unlimited recordings on its free plan, so this step costs nothing to complete. The two prompts take about forty minutes including the citation checks, and the citation checks are not optional, because every downstream asset inherits whatever this step claims.

This step converts research into the short set of decisions every campaign asset will repeat: the goal, the concept, the segments, the channels, and the pillars. CoSchedule found that goal setters are 377 percent more likely to report success than marketers who skip goals, so the measurable objective is not optional bureaucracy, it is the highest return sentence in the whole plan. Open ChatGPT, paste in the research summary from Step 1, and use this prompt:

Here is my campaign research summary. Act as a marketing strategist. Propose one campaign concept with a name, the single behavior I want the audience to take, and the objective expressed as one measurable goal, such as 500 trial signups in 21 days. Then give me: three audience segments ranked by fit, two messaging pillars per top segment written in the language my research uses, the three best channels for this campaign with a one sentence reason for each, a rough budget split across those channels, and the three to five numbers I should track to judge success. Mark anything that sounds generic so I can sharpen it.

[paste your research summary here]

Push back on the output before accepting it. Ask which claim a skeptical customer would challenge first, request a version of the concept that removes the word your closest competitor also uses, and cut the channel list to three, because HubSpot data showing 52 percent of teams on five to eight channels is a description of the market, not a recommendation for a first campaign. This iteration is where ChatGPT earns its 4.7/5 rating in our database, because strategy is a conversation rather than a one shot generation, and the free tier handles it with Go at $8 per month and Plus at $20 per month removing the usage caps free plans apply at peak hours.

Then stress test the messaging in Claude, which sustains long analysis better than any other model at 4.6/5 in our database, with Pro at $20 per month or $17 billed yearly. Create a Project for the campaign, paste in the pillars, and ask it to argue against you:

Here is my campaign strategy and messaging pillars. Act as a skeptical customer seeing this brand for the first time. List every claim you do not believe, every phrase that sounds like generic marketing, and the one question you would ask before signing up. Then rewrite the two weakest pillars in the language a real customer would actually use.

Lock four artifacts before moving on: the measurable goal, the campaign concept, the ranked channel shortlist, and the final two pillars. Save them in Notion AI as the campaign hub, because every later step references them, and paste the pillars at the top of every prompt you run from here on. That verbatim repetition habit is the cheapest way to make five tools sound like one brand.

This step produces the copy set the whole campaign runs on: one landing page, six social posts, and the announcement email. Start in Jasper, the enterprise content platform rated 4.3/5 in our database, whose Brand Voice feature stores your pillars once and applies them to everything it writes. Paste your strategy artifacts into this prompt:

Write the copy set for my campaign using the strategy and pillars I pasted. One: a landing page hero with a headline of ten words or fewer and a subhead of twenty words or fewer. Two: three LinkedIn posts of 150 words or fewer each. Three: three Instagram captions where the first line works as a hook. Four: one announcement email of 150 words with a subject line under 45 characters. Reuse my two messaging pillars in the same words in every asset, keep every claim traceable to the facts I provided, and no exclamation marks anywhere.

[paste your pillars, offer details, and proof points here]

Jasper has no free plan, only a 7-day trial with Pro at $69 per month or $59 billed yearly, so if budget is tight, run the same prompt in ChatGPT or Claude instead and keep Jasper as the upgrade when volume grows or teammates need shared Brand Voice controls. That trade is exactly what the free tier path in this guide is built around.

Then move to Anyword for the paid ad variants, because it is the one copy tool in our database that predicts performance before you spend: it attaches a score to every variation, which turns copywriting from taste into testing. Rated 4.3/5 with Starter at $49 per month and Data-Driven at $99 per month, it earns its price the moment real ad budget is behind the copy:

Generate five variations of this ad copy for my campaign. For each variation, show the predicted performance score and explain in one line why it scores the way it does. Keep the tone consistent with my messaging pillars, keep the primary text under 90 characters, and vary the opening hook rather than the offer.

[paste your winning copy and audience description here]

Close the step with a consistency pass in Grammarly, rated 4.4/5 with Pro at $12 per month billed yearly: check the whole copy set in one session so the suggestions treat it as one voice, and accept the tone suggestions only where they agree with your pillars, because Grammarly does not know your campaign goal and you do.

This step produces the two visual deliverables every campaign needs: the briefing deck that aligns everyone before launch, and the graphics that fill the channel calendar. Start in Gamma, the fastest prompt to deck tool in our database at 4.6/5, and paste your strategy materials into this prompt:

Create a 7 slide campaign briefing deck from my strategy materials. Slide one states the campaign objective and the single target behavior. Slide two profiles the primary audience segment in their own words from my research. Slides three and four show the two messaging pillars with the proof points behind them. Slide five maps the three channel choices with start dates. Slide six shows the budget split. Slide seven lists what each owner delivers and by when. Keep each slide under 40 words and suggest one simple chart or image per slide.

[paste your strategy summary and channel plan here]

Gamma generates the deck in one pass and the one click restyle feature lets you test themes before committing, which is why it earns 4.6/5. It is free to try with 400 credits, with Plus at $9 per month and Pro at $18 per month, and the Plus tier also removes the made with Gamma branding, which matters the moment stakeholders outside the team will open the file.

For the channel graphics, open Canva AI and use Magic Design to produce the post templates sized for each platform, then save the campaign colors and fonts in the Brand Kit so every future graphic matches without re-explaining the style. Canva AI holds 4.4/5 in our database, with a free tier, Pro at $18 per month or from $12 billed yearly, and Business at $25 per user per month. Two honesty rules keep the assets credible: screenshot real product screens rather than asking AI to invent them, because audiences forgive plain screenshots and punish fake looking renders, and pair every graphic with one line from your messaging pillars so the visuals and the copy tell the same story.

This step wires the campaign together so launch week runs on automation instead of adrenaline. Start with the social calendar in Buffer, rated 4.6/5 and the most small team friendly scheduler in our database: its free plan covers three channels with ten scheduled posts each, and the AI Assistant drafts and repurposes posts from one idea into channel specific versions. Feed it the six posts from Step 3 plus the pillar lines, let it queue them across your three channels, and schedule them against the dates in your briefing deck. When the campaign grows past the free caps, Essentials costs $5 per channel per month billed yearly, which keeps it the cheapest meaningful upgrade in the stack.

Then set up the email side in Mailchimp AI, rated 4.1/5: import the announcement email from Step 3, let the Content Optimizer score the subject line before send, and turn on send time optimization so each subscriber receives it at their hour of highest open probability. Mailchimp is free up to the list limit with Essentials at $13 per month, and the upgrade is worth it the moment you want the per subscriber reporting that Step 6 depends on.

Finally, describe the handoffs to Zapier Copilot in plain language and it builds the multi step Zaps for you:

Build these Zaps for my campaign. One: when someone submits my landing page form, add them to my Mailchimp audience with the tag campaign-spring and post a notification to my marketing Slack channel. Two: when someone completes the target action, tag them as converted in Mailchimp and add a row to a Zapier Table with the source and the date. Three: every Monday at 8 am, post the campaign performance summary to Slack. Then show me how to test each one safely before launch.

Test every Zap with fake submissions the day before launch, because a broken handoff discovered mid campaign is a campaign problem. Zapier connects more than 8,000 apps and earns 4.7/5 in our database, with a free plan covering 100 tasks per month and Professional at $19.99 per month billed yearly. If you outgrow it, Make at 4.5/5 is the visual alternative, with a free 1,000 operations per month and Core from $9 per month, though its learning curve is steeper. The automation principle for both: move data automatically, but leave the budget decisions and the customer replies to humans.

This step turns campaign week from a blur into a decision base. Before launch, write down the three to five numbers that define success, exactly as the Step 2 prompt produced them: cost per acquisition, conversion rate, and revenue by channel are the usual three. From launch day onward, collect the daily numbers in one place. HubSpot AI, rated 4.3/5, is the strongest home for this because the CRM captures every lead with its source automatically, and its free tier is genuinely usable, with Starter at $20 per month adding the reporting depth a multi week campaign needs. If your campaign runs outside a CRM, export the numbers to a spreadsheet and use ChatGPT with data analysis turned on instead:

I uploaded a spreadsheet of campaign data: daily spend, clicks, signups, conversions, and revenue, broken down by channel. Compute cost per acquisition and conversion rate per channel, rank the channels by return, chart cumulative conversions against my target of 500 in 21 days, and flag any day where a channel spent money without producing conversions. Summarize the three decisions I should make this week.

Change one variable at a time based on what the numbers say: a new hook in the paid ads, a different subject line in the email, or budget moved from the weakest channel to the strongest, never three changes at once, because then you cannot tell which one worked. Omnisend data from 2025 shows why patience pays on the email side specifically: automated emails generated $2.87 in revenue per email against $0.18 for regular campaigns, so the flow you built in Step 5 gets better as it accumulates history, and judging it after one send undervalues it.

Close the loop with a structured retrospective within a week of the end date, run in Notion AI: what the numbers said, which asset overperformed, what broke, and the three changes for the next campaign. Ask Claude to draft the retro summary from your notes so it is shareable with stakeholders. Then update the campaign hub so the next campaign starts from evidence instead of memory, which is precisely the documented strategy habit the CoSchedule data says separates successful marketers from the rest.

The six steps fit a two week part-time schedule, and the sequencing rule is simple: never start a step before the previous artifact exists, because every step inherits the quality of the one before it. Week one covers thinking and writing: the two Perplexity research prompts on day one, the ChatGPT strategy session and the Claude stress test on day three, and the full copy set in Jasper plus the Grammarly pass by the end of the week. Week two covers production and machinery: the Gamma deck and Canva templates on day one, the Buffer queue, the Mailchimp flow, and the Zap tests on day three, a full dry run of everything the day before launch, and the campaign goes live on schedule while you watch the first numbers arrive.

If you are the kind of planner who wants the calendar to enforce this, Motion at 4.5/5 auto-schedules the six sessions as tasks around your existing meetings, with Pro AI at $19 per seat per month after the trial and no free tier, which is honest friction: it is the one tool in this guide you pay for before you have proven the workflow. The lighter option is a simple board in Notion AI, where the campaign hub already lives, with one card per session and the artifact each session must produce written on the card. Either way, put the dry run on the calendar the moment you schedule the launch date, because it is the first thing a busy week eats and the last thing you want to lose.

The difference between a noisy campaign and a coherent one comes down to workflow habits. These seven come from the patterns that separate campaigns that convert from campaigns that merely publish.

• Give each tool one job. Perplexity researches, ChatGPT strategizes, Jasper writes at scale, Anyword predicts, Buffer schedules, Zapier automates. Mixing jobs produces mediocre results at both, while a single purpose chain lets each tool do what its rating in our database was earned for.
• Reuse the pillars verbatim. Paste the two messaging pillars at the top of every prompt in Steps 3 through 5. A campaign feels coherent when a visitor reads the same claim on the landing page, in the email, and on the social graphic, and verbatim repetition is the cheapest way to buy that effect.
• Declare the success numbers before launch. CoSchedule found goal setters are 377 percent more likely to report success, and the declaration matters more than the specific numbers: three to five metrics written down before the first ad spends a dollar, so the post launch conversation is about evidence instead of mood.
• Keep a prompt library in the campaign hub. Store every prompt that produced good output in Notion AI with a note on what it produced. The second campaign starts from proven prompts instead of the blank page, and the library becomes the documented strategy the data says most marketers still lack.
• Ask AI to attack the plan. Before launch, give ChatGPT the whole plan and ask it to act as a skeptical customer seeing the campaign cold, then list every moment of confusion. This five minute red team routinely surfaces the ambiguous claim a real visitor would bounce on.
• Repurpose from the pillar, not from the post. When a channel needs more volume, go back to the pillars and generate fresh, rather than recycling the same post across networks. The AI Assistant in Buffer adapts one idea per channel, which reads native instead of copy pasted.
• Timebox the polishing. The campaign exists to start conversations with customers, not to win a design award. One review pass per asset plus the dry run is enough; beyond that, you are delaying the market signal the campaign exists to produce.

Most AI planned campaigns fail in predictable ways, and every failure mode has a simple fix. Five mistakes account for the large majority of weak campaigns.

• Running the campaign without a documented plan. Almost 40 percent of marketers operate with no documented strategy, and AI makes that failure cheaper to commit, because generating assets is so fast that planning feels optional. The fix is structural: the workflow produces the documented plan as the byproduct of Steps 1 and 2, so refusing to skip them is the whole discipline.
• Spreading across every channel at once. The market average is five to eight channels per team, but a first campaign on five channels produces five shallow presences and no learning. The fix is the Step 2 shortlist: three channels, each with a one sentence reason, and one new channel tested per quarter after that.
• Letting AI invent claims the product cannot back. AI fills gaps with plausible sounding promises, and one unsupported claim in a paid ad costs more than the campaign. The rule from Step 3 holds everywhere: the AI writes only from facts you supplied, every market claim names its source, and the Perplexity citations from Step 1 are the audit trail.
• Skipping the dry run. Untested Zaps, unscheduled posts, and unproofed emails surface their bugs at the worst possible hour. The thirty minute rehearsal the day before launch is the highest return half hour in the entire workflow, and it is the first thing a busy week tries to eat.
• Measuring everything and deciding nothing. A dashboard of twenty metrics produces no next step, and campaigns without pre declared numbers get judged by mood. Declare three to five numbers before launch, read them weekly in HubSpot AI, change one variable at a time, and let the retrospective pick three changes, not thirty.

The table below summarizes the full campaign stack so you can see the whole workflow at a glance. Twelve of the fourteen tools offer a genuine free tier, and the complete paid stack would cost roughly $200 per month at list prices, though most teams pay for only one or two subscriptions during a campaign cycle and buy the rest against bottlenecks as they appear.

Free tiers carry the entire workflow in this guide, and every upgrade has a specific moment when it becomes worth buying: the moment its constraint actually interrupts the campaign. Buying in that order, instead of subscribing to everything up front, keeps your costs matched to your progress, which matters when 73 percent of marketers report their budgets are more heavily scrutinized than before.

• Zapier Professional at $19.99 per month billed yearly removes the 100 task monthly cap, which a live campaign burns through quickly once every form submission and conversion triggers a Zap. This is usually the first upgrade, because automation breaks exactly when campaign traffic arrives.
• Jasper Pro at $69 per month, or $59 billed yearly is the first paid writing tool worth it, and the trigger is volume or teammates: when the copy set grows past one landing page and three emails, or when several writers need the same Brand Voice, the trial has proven the workflow and the upgrade pays it back in hours saved.
• Buffer Essentials at $5 per channel per month billed yearly lifts the 10 scheduled posts per channel cap and adds deeper analytics. Upgrade when the content calendar for one campaign exceeds what three queues of ten can hold, which most multi week campaigns do in week two.
• Mailchimp Essentials at $13 per month becomes necessary once the audience passes the free list limit, and it unlocks the per subscriber reporting that Step 6 depends on for the email verdict.
• ChatGPT Go at $8 or Plus at $20 per month removes the usage caps free plans apply at peak hours. Upgrade when a strategy session gets interrupted mid conversation, ideally in week one rather than mid crisis.
• Canva Pro at $18 per month or $12 billed yearly expands the AI generation allowance and unlocks the full Brand Kit, which matters the moment the campaign produces more than a handful of graphics per week.

For context, the full core stack costs about $200 per month at list prices if everything is paid at once, and that is rarely necessary. The practical pattern is one or two subscriptions at a time, bought against the bottleneck you are actually feeling, and revisited after the campaign when the workflow changes shape. The two specialists sit outside that budget by design: Semrush Copilot at $139.95 per month and Anyword at $49 per month earn their place when competitive intelligence and paid ad testing become recurring needs rather than one campaign experiments.

Most teams treat each campaign as a fresh start, and that is exactly why the second campaign costs almost as much as the first. The same AI workflow that planned this campaign becomes a system when its artifacts persist: the research document, the pillars, the prompt library, and the retrospective live in the campaign hub in Notion AI, and the next campaign starts by updating them instead of recreating them.

Run a quarterly cadence between campaigns. Every month, review the channel numbers in HubSpot AI and pick one insight worth acting on, because the CRM accumulates lead sources whether you look at them or not. Every quarter, test exactly one new channel with a small budget, since HubSpot found 94 percent of teams diversified their channels last year and the teams that do it deliberately beat the teams that do it reactively. Between tests, ask Claude for a fresh angle on the winning pillar, which gives proven messaging a second life without inventing new claims, and feed the customer questions that actually arrived into ChatGPT to update the FAQ and objection handling in the copy set.

The compounding effect is the real story. CoSchedule research found AI saves marketers more than five hours per week on average, and a documented workflow multiplies that saving because the second campaign reuses the prompts, the templates, and the automation from the first. By the third campaign, most of the work shifts from creating assets to choosing between proven ones, and that is the position a small team wants to hold when a competitor is still starting from a blank page.

Reading about a workflow does not run a campaign, so here is the whole guide compressed into a checklist you can start today. Each block maps to a step, with the realistic time it takes on free tiers across a two week part-time schedule.

• Session 1, about 2 hours: run the two Perplexity prompts from Step 1, check the citations, record one customer call with tl;dv if you can, and save everything into one research document.
• Session 2, about 1 hour: run the ChatGPT strategy prompt from Step 2, push back on the generic parts, and lock the goal, concept, channel shortlist, and two pillars into the Notion campaign hub.
• Session 3, about 2 hours: run the Claude stress test, then write the copy set with the Jasper prompt and run the whole thing through the Grammarly pass from Step 3.
• Session 4, about 2 hours: generate the briefing deck in Gamma, produce the channel templates in Canva AI, and pair every asset with one pillar line from Step 4.
• Session 5, about 2 hours: queue the social calendar in Buffer, build the email flow in Mailchimp AI, and build and test the three Zaps with Zapier Copilot from Step 5.
• Session 6, campaign weeks: rehearse everything the day before, publish on the schedule, read the weekly numbers in HubSpot AI with the Step 6 prompt, and hold the retrospective within a week of the end date.

Six sessions produce a campaign that reads as coordinated, spends against a documented plan, and leaves you with the data to do it again better. Start with the first Perplexity prompt this morning, and by the end of the week the campaign stops being a scramble and becomes a process.