Field guide · Paid acquisition
Your first AI ad campaign.
Use AI to make campaign research and creative iteration faster, not to skip the decisions that make paid acquisition work. Start with the offer and measurement, then build a small test you can actually learn from.
Campaign architecture
Six decisions before Ads Manager.
The interface will change. The campaign logic is more durable. Make these decisions before optimizing settings inside any one ad platform.
Goal
Choose the business action you need: purchase, qualified lead, install, signup, trial, or another measurable event.
Audience
Define the situation and customer clearly enough that the creative can speak to something specific.
Offer
Make the next step understandable. What does the person get, why is it useful, and what happens after the click?
Hypotheses
Write several reasons the customer might care, not several rewrites of the same headline.
Creative
Turn each hypothesis into an ad that can communicate the idea quickly and credibly.
Measurement
Track what happens after the click so you do not optimize only for cheap traffic.
Brief
Give AI the truth before asking for copy.
A good model can multiply the quality of your inputs. It can also turn vague assumptions into polished nonsense. Build the brief from source material first.
Customer
Who is this version for? Describe the person, situation, and existing alternative.
Problem
What painful or valuable job is already on their mind?
Promise
What outcome can the product credibly create?
Proof
What demonstration, result, screenshot, review, mechanism, or evidence supports that promise?
Offer
What exactly are you asking the person to try, buy, book, install, or start?
Objection
What is the strongest reason they might not believe or act on the ad?
Creative hypotheses
Write different reasons to care, not three synonyms.
Pain
The current problem is frustrating enough that relief creates attention.
Outcome
The desired result is more motivating than the pain itself.
Demonstration
Seeing the product work is more persuasive than explaining it.
Comparison
The customer needs to see why this is different from the current alternative.
Objection
Addressing skepticism directly may unlock the next step.
AI workflow
Use the model as a creative partner, not a source of facts.
Give the model the real product, customer research, offer, constraints, and proof available.
Ask for distinct creative hypotheses before asking for complete ad copy.
For each hypothesis, generate several hooks that express the same core idea differently.
Choose the hooks a human can defend. Remove claims that require invented data or testimonials.
Generate copy, visual briefs, shot lists, or asset variants around the selected concepts.
Review the final ad as a customer would see it, not as a prompt-writing exercise.
A useful prompt starts with source material and ends with a constraint: "Use only claims supported by the information above. Flag anything you cannot verify instead of filling the gap."
Before launch
Make the test diagnosable.
Tracking
Confirm the primary conversion event and any important downstream event are recording correctly.
Naming
Label campaigns, ad sets, and creatives so you can tell which hypothesis, audience, and version produced the result.
Destination
Check the landing page or app-store experience on mobile. The message should continue after the click.
Creative set
Launch enough meaningful variation to learn, but not so many changes that you cannot explain the result.
Guardrails
Set a budget and stop conditions you can afford before launch. Do not invent a universal budget or kill rule.
Read the result
Follow the customer beyond the click.
Attention
Did the right people stop, watch, or engage with the idea?
Click
Did the creative create enough intent to take the next step?
Conversion
Did the destination and offer turn that intent into the primary action?
Activation
Did new users or leads reach the first meaningful outcome?
Quality
Did the campaign produce retained users, qualified leads, revenue, or another downstream signal that matters?
Diagnosis
The pattern tells you where to look next.
Strong attention, weak clicks
The creative may be entertaining without creating product intent, or the CTA may not follow naturally from the story.
Strong clicks, weak conversion
The ad may be overpromising, the destination may be unclear, or the offer may not be strong enough.
Strong conversion, weak activation
Acquisition is working better than onboarding or the ad is attracting the wrong expectation.
Strong activation, poor retention
The product may solve a one-time problem, fail to create a habit, or attract users whose need is not recurring.
Good downstream quality, expensive acquisition
The message may be right while creative efficiency, audience reach, or conversion still needs work.
Seven days
Build one campaign you can learn from.
Define the business goal, customer, offer, conversion event, and the evidence available.
Write five creative hypotheses and choose the three that are meaningfully different.
Use AI to expand hooks, scripts, briefs, and visual directions for those three hypotheses.
Produce the assets and review every claim, CTA, crop, caption, and destination on mobile.
Configure tracking, naming, audience, budget guardrails, and the destination experience.
Launch the controlled test and record what each creative is meant to teach.
Read the funnel, write the diagnosis, and choose the next variable to change.
Keep going
Turn the first campaign into a repeatable system.
Build alongside
other marketers.
Ask questions, share what you're building, and get unstuck faster. The community moves at the pace of people who are actually shipping.
Ask anything
No question too basic
Share your builds
Real feedback from peers
Early access
New guides first
Free to join · No credit card · Unsubscribe any time