Playbooks
43 published pages on playbooks for operators working through real growth constraints.
Turn Curated Growth Ideas Into Testable Experiments
Start with a current growth decision, then select individual posts that suggest a relevant mechanism. Trace each claim to its source, separate observed results from explanations, and write a falsifiable experiment brief. Treat curation as a discovery aid; measure whether your process produces useful tests without consuming too much research time.
How to Test Whether an AI Agent Interface Actually Saves Time
Compare the new interface with your current workflow on comparable tasks while keeping the underlying agent and prompt intent consistent. Measure time through review and final placement, alongside acceptance rate, revisions, and context failures. Include failed attempts and recovery time before deciding whether the interface saves work.
Build an Article Illustration Workflow That Explains One Idea at a Time
Start with the idea a reader needs to understand, then give each illustration one explanatory job. Write a visual brief, choose a simple metaphor, generate the image, and check its accuracy and readability. Before making illustrations standard, compare an illustrated article with a text-only version using the same passage as the completion checkpoint and clearly defined comprehension measures.
Order Confirmation Emails: Help Customers Know What Happens Next
An order confirmation email can begin onboarding by explaining what the customer bought, what happens next, and how to prepare for first use. Keep transaction details easy to find, then add a relevant benefit recap, realistic expectations, and one useful next step. Test downstream behavior before claiming the message improves retention.
How to Test a One-Screen Free-Trial Paywall
A one-screen free-trial paywall puts the trial benefit, relevant proof, action button, and billing terms together at the decision point. Test it against your current paywall while keeping price, trial duration, and onboarding constant. Judge the result by paid subscribers per eligible viewer after a complete billing observation window, with refunds and retained economics as guardrails.
Apple Ads Keyword Decisions: Connect Spend to Subscription Margin
Evaluate Apple Ads keywords by connecting acquisition spend to subscription cohorts at the most detailed level your attribution supports. Compare cohorts at the same age, subtract refunds and relevant variable costs, and keep unmatched records visible. Set evidence thresholds and spending limits before deciding to hold, cut, or scale; attributed revenue alone does not establish incremental growth.
Cold Email for Startup Jobs: Build a Measurable Outreach Test
Start with a small set of startups where your experience matches an observable need. Send the relevant hiring decision-maker a short email with a specific reason for reaching out, one or two credible proofs, and one clear ask. Compare applications alone with applications plus email, then track qualified hiring conversations and time spent. Treat the results as directional evidence, especially with a small target list.
How to Test Whether AI Video Production Actually Saves Time
Compare your existing video workflow with an AI-assisted workflow using the same representative briefs and a shared quality rubric. Count all production and revision effort, including time spent on rejected outputs, and separate hands-on work from elapsed turnaround. Test audience performance separately: producing usable videos faster does not establish that they generate better business results.
How to Test a Job Application Agent Before Giving It More Autonomy
Start with a small pilot that compares manual applications with an agent workflow requiring review before every submission. Use the same résumé and qualification rules, measure active time per confirmed application, and track material corrections across all reviewed applications. Keep sensitive answers under human control, and treat recruiter-screen outcomes as directional evidence until the sample and follow-up are sufficient.
AI Interview Practice for Growth Leaders: Build a Useful Feedback Loop
Start with a bank of verified career stories, then record answers to role-relevant questions and review them against a consistent rubric. Use AI to identify a specific weakness and suggest a focused drill. Test progress on unfamiliar questions with a human reviewer; rising AI scores alone do not establish better interview performance.
How to Compare AI Models by Cost per Usable Output
Compare candidate models on the same documents, prompts, and output requirements, then judge their outputs against a predefined quality threshold. For each candidate, divide its workflow cost—including failed attempts, retries, and required review—by its accepted outputs. Track the cost of running the evaluation separately from the recurring cost of operating the workflow.
How to Test Persistent Memory for a Growth Agent
Test persistent memory on recurring tasks that require an agent to apply earlier decisions, corrections, and definitions. Compare no persistent memory, a searchable log, and hierarchical summaries using the same history and task conditions. Score correct application of history alongside stale recall, unsupported claims, token usage, latency, and maintenance effort.
How to Test Whether AI Agents Can Use Your Website
Test agent readability by giving independent AI agents specific website tasks with predefined correct answers and completion criteria. Establish a baseline, fix a small set of observed failures, and repeat the same tasks against changed and unchanged pages. Use audit scores to generate hypotheses, then judge fixes by correct task completion, human experience, and the business relevance of the task.
How to Benchmark AI Models for a Growth Workflow
Benchmark AI models against one defined growth workflow using the same task inputs, acceptance criteria, and scoring rubric. Compare usable outputs, total cost, review effort, latency, and failures across repeated runs. Before connecting a provider, review request routing and credential handling. Add routing only when the observed improvement justifies the extra operating burden.
How to Test Whether More Ad Creative Actually Helps
Compare your current creative cadence with a higher cadence while keeping media budget, offers, eligibility, and bidding comparable. Define what counts as a distinct concept, measure customer outcomes over the same window, and include production costs. A comparison between cadences estimates the benefit of the operating change; measuring advertising’s total incrementality requires a separate no-ad holdout.
How to Choose Growth Workflows for AI Agents
Choose workflows with a recurring trigger, usable inputs, bounded tools, a measurable finish line, and decisions that depend on the case. Then assess business value, error cost, and whether simpler automation could do the job. Test a narrow candidate alongside the existing process before giving it authority to change live systems.
Before You Pause a Meta Ad, Test Its Effect on the Creative Portfolio
An ad's attributed CPA does not establish what would happen if you removed it. Compare a stable conversion-focused creative with a portfolio that adds predefined support creatives, and judge total acquisition outcomes across comparable groups. Treat improved attribution as a clue; claiming incremental value requires evidence of additional business outcomes against a credible counterfactual.
How to Test a Scroll-Linked Landing Page
Test a scroll-linked landing page on one bounded campaign with a clear offer and an accessible path to purchase. Randomly split eligible visitors between the immersive page and a conventional page with the same message, products, and offer. Judge the treatment by completed purchases and contribution margin per visitor, while checking mobile usability, performance, and production cost.
When to Ask for Signup: Test Value Before Account Creation
Ask for signup when an account enables a useful next step, such as saving progress or continuing an experience. If users can safely experience value first, test moving mandatory account creation after that moment. Compare activation and retained use across everyone assigned to each flow, including people who abandon onboarding.
How to Evaluate AI Video Production Before Scaling Creative
Compare the AI workflow with your current production process using the same brief, assets, formats, and acceptance criteria. Measure total human hours per approved creative, including failed attempts and revisions, alongside elapsed time and production costs. Review quality without revealing the production method where practical, then test campaign performance separately. Faster production is useful only when the resulting videos meet your standards and test meaningfully different ideas.
How to Compare an API-Based SEO Stack With Your Current Suite
Run one recurring SEO task through both systems with the same brief and quality rubric. Compare total operating cost per approved analysis, including API usage, analyst time, maintenance, and failures. Replace that workflow only if the alternative meets your quality requirements and the savings survive realistic usage volumes.
How to Test Whether Shoppable Creator Content Adds Profitable Sales
Compare native product tags with conventional links using randomized assignment where feasible, consistent offers, and a defined purchase window. Evaluate the difference in contribution after campaign costs, using retailer order data where available. A tags-versus-links test estimates the value of the shopping interface; measuring the value of the entire creator campaign requires a separate control without that campaign.
How to Test Audience Participation in AI Livestreams
To test audience participation in an AI livestream, connect viewer submissions to visible changes in what plays next, then compare that experience with passive viewing. Measure active watch time, return visits, and the full path from submission to playback. Treat selection rules, generation delays, moderation, and cost as product decisions that can strengthen or break the experience.
How to Test Whether Audience Control Brings Livestream Viewers Back
Compare an interactive livestream with a passive version, assign visitors consistently, and measure return visits across everyone assigned. Instrument the path from submitting an idea to seeing its payoff so you can distinguish a broken interaction from a weak retention hypothesis. Keep content exposure comparable and set cost, latency, and moderation limits before interpreting any lift.
How to Test Meta Value Rules Against Broad Targeting
Compare a broad-targeting control with a treatment that adds one value rule grounded in customer economics. Keep the other campaign settings consistent and evaluate customers over the same contribution-margin window. Treat the multiplier as an experimental setting; better attributed results alone do not establish incremental profit.
How to Evaluate AI-Generated Interfaces for Growth Experiments
Give the generated-interface workflow and your current workflow the same bounded brief, then compare the time each takes to produce an acceptable interactive prototype. Include revision work and use the same task-based review for both. Treat production experiment readiness as a separate decision: faster prototyping does not establish reliable measurement or better conversion.
Build an AI Creative Workflow That Produces Useful Tests
Start with one customer dataset, one channel, and a clear definition of a useful creative test. Connect each brief to source evidence, generate variants around an explicit hypothesis, and keep launch and budget decisions with a named operator. Evaluate completed tests and total human effort alongside downstream conversion costs; asset volume alone does not show that the workflow helps.
How to Benchmark AI Creative Tools Before Adding Them to Production
Compare the candidate tool and your current workflow using the same campaign brief, references, deliverables, and time budget. Measure approved, consistent assets per production hour, including retries and corrections. Then test whether approved assets improve campaign performance; production efficiency and advertising effectiveness are separate decisions.
A Paid Media Holdout Protocol for Better Budget Decisions
To evaluate whether a paid channel earns its budget, compare normal exposure with a concurrent holdout and estimate the difference in business outcomes. Define contribution costs, the customer population, the measurement horizon, and decision thresholds before launch. Make the budget decision from incremental contribution and its uncertainty; use incremental CPA when the incremental acquisition estimate is positive and sufficiently precise.
How to Test a Growth Agent Before Giving It Control
Start with one recurring growth review and give the agent read-only access to the evidence needed for that decision. Define its objective, guardrails, decision history, and proposal format, then compare its recommendations with manual analysis using the same observation window. Expand its permissions only when the pilot demonstrates useful decisions at an acceptable total cost; evaluate business impact separately through experiments.
An SEO Content Refresh Strategy for Choosing What to Work on Next
Prioritize SEO refreshes where existing search demand overlaps with a specific, fixable problem and a useful business outcome. Route each opportunity to a refresh, consolidation, technical investigation, or new page before assigning editorial work. Test a comparable group of refreshes against a delayed-refresh holdout, measuring qualified organic traffic per editorial hour alongside downstream value.
Which Ads Should You Kill or Scale? Build a Spend–CPA Decision Map
Plot each ad’s spend against its attributed CPA, then compare it with an economically justified target and stopping threshold. Before acting, check conversion volume, reporting delays, and whether the ads are comparable. Treat efficient ads with sufficient evidence as scale candidates, uncertain ads as bounded tests, and persistently inefficient ads as stop candidates.
How to Evaluate an AI Marketing Workflow Before You Adopt It
Compare the AI workflow with your current process on comparable tasks using the same source evidence and acceptance rubric. Measure total human minutes per independently accepted deliverable, including preparation, review, revisions, and effort spent on failed tasks. Use the workflow's own quality scores as diagnostic suggestions and check them against independent judgments. Any claim of improved campaign performance requires a separate live test.
How to Test an AI Ad Production Workflow
Compare the proposed AI ad workflow with your current process using matched creative briefs and the same approval criteria. Measure total production cost per approved ad, including rejected outputs, retries, editing, and human review. Treat production efficiency and advertising effectiveness as separate questions: cheaper approved creative still needs a media test.
Product-Page Optimization: Find and Test Purchase Uncertainty
Start product-page optimization by identifying what shoppers need to know before they can buy: the product’s value, credible proof, decision-critical details, and the next purchase action. Use customer questions and behavior to choose changes, then test a controlled variant. Judge the result by completed orders and contribution margin per visitor, with add-to-cart behavior as a diagnostic.
When Is an AI-Built Growth Tool Worth Building?
An AI-built growth tool is worth pursuing when its expected recurring value justifies specification, implementation, review, integration, adoption, and maintenance. Test that proposition with a narrow workflow, predefined acceptance criteria, and a comparison against the current process. Measure total labor per accepted tool still in use, including effort spent on failed attempts.
How to Evaluate Managed Agents for Growth Reporting
Evaluate managed agents by running the same fixed reporting tasks through managed execution and your existing implementation. Define acceptable analysis before testing, review outputs without revealing which implementation produced them, and measure total engineering minutes per accepted output. Include setup, debugging, and recovery effort, then require both options to meet predefined quality, completion, cost, and latency limits.
How to Plan a Geo-Lift Test for a Meta Budget Decision
A geo-lift test compares business outcomes in geographic areas where a specific advertising tactic continues with areas where it is withheld. Start with the budget decision, then define comparable markets, assignment, spending contrast, outcome, and analysis before launch. Use the estimated incremental customers and corresponding spend difference to assess acquisition economics, while reporting uncertainty and limiting conclusions to the conditions tested.
How to Simplify Growth Operations Without Losing Useful Checks
To simplify growth operations, map one recurring workflow and identify a step whose function is already covered or no longer needed. Test its removal on a narrow set of eligible work, with a named owner, quality checks, and a rollback condition. Measure delivery time alongside errors and rework before deciding whether to retire the step.
How to Test Creative Growth Ideas Beyond Shares and Traffic
Start with a recognizable audience truth, express it through an unexpected format, and give people a relevant next step toward the product. Test that experience against a conventional campaign and, where feasible, a holdout receiving no additional campaign. Judge the result on incremental commercial value after costs, with sharing and traffic used to explain what happened.
How to Calculate Customer Acquisition Cost Without Fooling Yourself
To calculate customer acquisition cost, divide the acquisition costs for a defined period by the new paying customers acquired in that same period. The difficult part is choosing the costs and customers that match the decision. Platform CAC, blended marketing CAC, and fully loaded CAC answer different questions. Label the version, time window, attribution rule, and customer definition every time you report it.
Customer Research for Marketing: Turn Interviews Into Growth Work
Customer research for marketing is useful when it changes a specific decision: which problem to lead with, which proof to show, how to explain the product, when to contact a customer, or what to test next. Start with a decision, collect language and behavior from several sources, code the evidence by context and outcome, and turn recurring patterns into briefs—not a folder of memorable quotes.
Feature Adoption Strategy: What to Do After the Product Ships
A feature adoption strategy defines who should use the feature, the problem it solves, the behavior that signals need, how eligible users will discover it, what first value looks like, and what should bring them back. Plan the message, trigger, channel, product path, measurement, and follow-up before launch. Shipping makes the feature available; adoption makes it useful.