# Start Here: Choose Your Product Growth Job > Pick the problem on your desk, then take the one route that helps you move it. - Author: Rishikesh Ranjan · Published: Sep 17, 2025 · Updated: Aug 29, 2026 - Type: Essay - Tags: GTM, Resources - Growth levers: Activation (primary), also Acquisition, Retention - ~2000 words --- Start with the problem on your desk, not a category label. If you are staring at a weak activation number, an AI sales pitch, or a retention chart that makes no sense, the useful first read is different in each case. This page is a selector, not an archive and not a promise that one tactic fits every company. Pick the route that matches the decision you need to make this week. Read it with your own product, segment, and time window in mind. Then leave with one measurement, one experiment, or one question you can put in front of the team. ## Pick a growth route before you browse Growth content gets unhelpful when it lets you consume five unrelated ideas and call that progress. You do not need another saved thread about a company you might study later. You need to know what the next hour of thinking is for: understanding a mechanism, deciding whether to run a pilot, establishing a baseline, calculating a number, or finding where the number breaks. Those jobs overlap, but they are not interchangeable. A company teardown can give you a transfer test for a product decision. A playbook can make a pilot smaller and safer. A benchmark can stop a team from comparing incompatible numbers. A calculator can settle the arithmetic. A completed raw-event cohort table can show when a measured loss begins, while a projection can make a Month 1 and later-month assumption visible. Choosing the wrong one is how a real question turns into a week of vague discussion. > **Use this hub as a current route map:** Each recommendation below is checked against a live local page and its canonical route. That makes this a route map for the site as it stands today, not a permanent ranking of every possible growth resource. There is a useful sequence here when the problem is retention. First calculate the period so everyone is discussing the same number. Then build or inspect a completed raw-event cohort table if you need to locate a measured drop. A three-assumption projection can help you state the Month 1 and later-month scenario you want to test, but it cannot locate the drop. Only after the table shows a pattern should you debate onboarding, lifecycle messages, pricing, or product work. Reversing that order feels faster because tactics are familiar. It also makes it easy to spend a month improving a part of the journey that was never responsible for the loss. The same rule applies to growth stories and AI tools. A company example can give you a hypothesis, but it cannot answer whether your users will value the same interaction. An AI SDR workflow can be impressive in a demo and still fail your funnel. Read for the decision you own, and keep a small note beside you: what would have to be true here for this route to change what we do? ## Study a growth mechanism you can test Choose this route when you have an activation or product-adoption question and want to study a mechanism before copying a tactic. A teardown is a mechanism-and-evidence analysis: it asks what a company did, what condition made that move work, and what would have to be true for your product to borrow the idea. That is a much higher bar than collecting a list of growth tricks. The [Wispr Flow growth teardown](https://www.productgrowth.blog/p/wispr-flow-growth-teardown) is the right first stop if you are trying to understand how a product can turn a useful interaction into a repeatable habit. Read it looking for the transfer test: where does the product remove friction, how does a user discover that value in context, and what trust or reliability boundary could stop the loop from working for you? Do not leave with a borrowed launch sequence. Write down one behaviour you want a new user to repeat, the moment when they could first see its value, and the product condition that could make the behaviour fail. If you cannot name those three things, you are still reading a story. If you can, you have the start of an activation experiment. > **Steal this:** Treat a teardown like a lab notebook, not a vending machine. The move worth borrowing is the one whose condition you can reproduce, and whose limit you can see before your users find it for you. ## Decide whether an AI SDR pilot belongs in your funnel Choose this route when someone has asked, "Should we buy an AI SDR?" That question is usually too large. The decision is closer to buy, pilot, or wait. A tool may be able to draft, research, sort, or follow up, while still being a poor fit for a sensitive segment, a messy CRM, or an outbound motion with no clear owner. The [AI SDR playbook](https://www.productgrowth.blog/p/the-ai-sdr-playbook-what-actually-works) is for turning that broad debate into a bounded pilot. Use it when you need to decide what work a human still owns, what baseline you will compare against, and what deliverability, compliance, or escalation guardrail ends the test. A pilot without those boundaries is procurement wearing an experiment costume. Make the pilot boring on purpose. Pick one segment, one message family, one handoff point, and one person who can stop it. Record the baseline before the tool touches anything. If the workflow cannot improve a named stage without creating a deliverability or quality problem elsewhere, waiting is a valid answer. ## Set a SaaS growth benchmark before chasing a target Choose this route when the argument in the room sounds like, "Is this number good?" A benchmark is a comparison with a defined population and method, not a number you pin to a dashboard because it feels ambitious. Before you compare anything, decide what the metric includes, which period it covers, and whether the companies in the comparison resemble your business enough for the result to teach you anything. Read the [SaaS growth benchmarks](https://www.productgrowth.blog/p/saas-growth-benchmarks) page when you need that comparison discipline before setting a target. It is not a scorecard that tells every SaaS company what to chase. Its useful job is narrower: help you select a metric, inspect the definition and scope behind a comparison, and notice when a segment or time frame makes a neat-looking number misleading. Start with the metric closest to the decision you can change. If you are planning an activation rework, use an activation measure before a revenue headline. If customers are leaving, start with retention before inventing a top-of-funnel campaign. The point is not to find a flattering target. It is to pick a number that can tell you what to do next. ## Calculate retention before choosing a fix Choose this route when you have customer counts for a period but do not yet know what they say. Retention answers a constrained question: of the customers you started with, how many remained after removing accounts active at the end that were absent at the start? It does not tell you why they stayed or left. It gives you the clean starting number for that investigation. Use the [retention rate calculator](https://www.productgrowth.blog/calculators/retention-rate) when you want the period math before you choose a retention tactic. Enter the starting customer count, accounts active at end that entered after start, and ending customer count for the same period. Keep the unit consistent too: customer logos, paid accounts, and seats can tell very different stories. The result is a measure, not a diagnosis. Take a simple operating case. You begin with 800 customers, 90 end-active accounts entered after the start boundary, and you end with 760. The calculation first removes those 90 accounts, then asks what remains of the original 800. That result is the conversation starter. To ask whether the loss arrived in the first month, after a pricing change, or in one acquisition channel, build a completed raw-event cohort table with consistent definitions. Do not jump from one blended result to a retention programme. > **A retention result needs a time frame:** A monthly customer-retention result and an annual revenue-retention result are not substitutes. Name the period and the thing being retained before you compare either one with a benchmark or a previous result. ## Use a raw cohort table when retention has a timing problem Choose this route when you already have a retention number and the next question is, "When does the leak begin?" A completed raw-event cohort table groups people by a shared start point, such as a signup month, under a fixed inclusion event, return event, and time rule. Comparing those completed groups over time can show whether a measured loss appears early or later. A blended rate cannot show that timing on its own, and the table still cannot prove the cause. The [cohort analysis tool](https://www.productgrowth.blog/tools/cohort-analysis) is the right next stop when you want to make a retention scenario explicit before you measure it. It is an illustrative three-assumption projection: cohort size, assumed Month 1 retention, and an assumed later-month multiplier. It does not read events, show an observed curve, or locate a measured drop. Use it to state the question or target you want to test, then build or inspect a completed raw-event cohort table in your own data. Only after a completed raw-event cohort table shows a measured pattern should you choose an early-versus-late follow-up. If the table shows most loss in the first month, inspect activation: the first meaningful value event that gives a new user a reason to return. If it shows returns holding through month one and then declining later, investigate the recurring job, the repeat-value journey, or a segment that looks healthy only in the average. The table tells you where to look, not why the pattern happened. ### Choose the next job, then do it You do not need to read every route before acting. Pick the one that matches the tension in front of you. If you are deciding what a product interaction could teach you, study the teardown. If an AI SDR proposal is waiting for approval, bound the pilot. If a dashboard argument has no shared definition, establish the benchmark. If retention looks wrong, calculate the period first. When timing matters, build or inspect a completed raw-event cohort table, and use the projection only to make its Month 1 and later-month assumptions explicit. My default starting point is measurement. It is harder to talk yourself into a tactic when the team has agreed on the metric, its definition, and the comparison that makes it useful. Start there, write down the question the number cannot answer, and let that unanswered question choose the next route. ## Common product growth route choices #### Which product growth route should I take first? Start with the decision in front of you. Study a teardown when you need a mechanism to test. Use the AI SDR playbook when you need to bound a buy, pilot, or wait decision. Start with benchmarks when the team has not agreed on the metric or comparison. Calculate retention when you need the period-level number, then build or inspect a completed raw-event cohort table when you need to know when a measured loss begins. #### When should I use retention rate versus cohort analysis? Use retention rate first when you need to calculate the share of starting customers who remained during one defined period. Use a completed raw-event cohort table when that result creates a timing question. The table groups people by a shared start point, such as signup month, under a fixed inclusion event, return event, and time rule, so you can see whether a measured weakness appears in the first experience or later. The cohort tool is only an illustrative three-assumption projection. Use it to state assumed Month 1 retention and a later-month multiplier, not to diagnose an observed curve. #### Should I start with a benchmark or a growth teardown? Start with a benchmark when you need to define a metric, a period, and a valid comparison before setting a target. Start with a teardown when you already have a product question and want to test whether another company's mechanism could transfer to your situation. A benchmark sets the frame. A teardown gives you a hypothesis. **Next job: Set a benchmark before you set a target.** Use the SaaS growth benchmarks guide to choose the metric that fits your decision, check the comparison behind it, and avoid setting a target that mixes incompatible segments or periods. [Continue](https://www.productgrowth.blog/p/saas-growth-benchmarks) --- All posts: https://www.productgrowth.blog/archive · Site: https://www.productgrowth.blog