Blog

Feature Gating Strategy: A 3-Step AI Workflow for Product Managers

Pricing and Monetization
Blog Author: Rina Alexin

Table of Contents

Feature gating means limiting or removing a feature in a lower pricing tier so customers upgrade to use it fully. A good feature gating strategy gates what grows with customer value, and leaves alone the features everyone already relies on. 

Most product managers don’t get much say in that decision. During our recent Pricing and Monetization webinar, I asked the audience who owns pricing at their company. The answers came back finance and the CEO or exec team. 

I think that leaves the best evidence out of the room. Product managers know which features customers rely on and which ones upgrading accounts adopt first. And pricing is a powerful lever. In Marn and Rosiello’s classic Harvard Business Review analysis, a 1% price improvement lifted operating profit by 11.1%, assuming sales volume held steady. 

In the webinar, Roger Snyder and I used AI to work through a packaging decision for Planloop, a fictional project management tool. Planloop wanted to lift quarterly free-to-paid conversion from 13.5% to 16% without hurting signups or churn. This post walks through the same workflow so you can run it on your own product. 

 

Why Usage Data Alone Picks the Wrong Features to Gate 

Feature usage is not willingness to pay. The most-used feature in a product is usually a core requirement, so gating it backfires. Gate features that larger or upgrading accounts use more, that competitors don’t give away free, and cap access rather than removing it entirely. 

Think about Zoom. Video calling is its most-used feature because it is the product. Put that behind a paywall and you lose users instead of gaining revenue. 

Planloop’s data showed the same trap. Comments and @mentions were used weekly by 93% to 97% of every account group, free and paid alike. Gate comments and you hit every free user without learning anything about who’s ready to pay. 

 

The Three Tests Every Feature Gating Decision Must Pass 

To decide which features to gate in a SaaS product: 

  1. Compare feature usage between accounts that upgraded and accounts that stayed free. 
  2. Confirm larger or growing accounts use the feature more as they scale. 
  3. Check competitor pricing pages so gating won’t make your free tier look weak. 
  4. Prefer a usage cap over a full lockout so free users still reach value. 
  5. Test the gate with a new-signup cohort, fake door or upgrade prompt before rollout. 

Those steps rest on three tests. I’ll use an invoicing feature inside a project management tool to show each one. 

Value grows with the account. An individual free lance project manager might send one or two invoices a month, then five or six as their business grows. Cap invoices on the free plan and the limit meets customers at the moment they’re getting more value. 

Upgraders use it more. Compare free accounts that upgraded with accounts that stayed free. If the free-forever accounts never touch invoicing, that’s a strong sign on which kinds of accounts to target with upgrade messaging. 

The market won’t punish you. Say a competitor offers unlimited invoices on its free plan. Gate invoicing and your product suddenly looks weak next to theirs. As Roger put it during the session, once several competitors set an expectation, you’re at a disadvantage if you don’t meet it. You can still use that feature as a signal for your sales team. It just may not be the right gate. 

I usually prefer limited access over a full lockout for some core features. Free users who never get to try a feature have no reason to pay for it. A cap lets them see the value first, and hitting the limit becomes a natural moment to upgrade. 

Here’s what Planloop’s usage data looked like. The figures show the share of accounts using each feature weekly over the last 90 days. 

Feature Free: upgraded Free: stayed free Pro Business 
Comments and @mentions 95% 93% 96% 97% 
Kanban boards 90% 88% 91% 86% 
Timeline view 71% 22% 58% 83% 
Guest access 54% 19% 41% 63% 
Integrations 48% 31% 66% 79% 

Planloop is a fictional company. The data is for practice only. 

Timeline view and guest access pass the first two tests easily: upgraders used them far more than accounts that stayed free. Integrations show a smaller gap but are worth a closer look. The third test needed competitive context, which is where AI came in. 

 

The 3-Step AI Workflow: Usage Data, Competitive Context, Experiments 

AI can take a PM from raw usage data to a testable packaging change in under an hour. It works in three steps, and you make the call at the end of each one. 

Step 1: Find the upgrade signals in your usage data 

Export account-level feature usage for free and paid accounts. Your product ops or operations team should be able to pull this for you. Then upload it to your AI assistant. 

The prompt matters more than you’d expect. Compare these two: 

Quick prompt: “Here’s our feature usage data. Which features should we move into a paid tier?” 

Better prompt: “Act as a pricing analyst. Using this usage data, find features where adoption rises with account value, and where free accounts that upgraded used the feature far more than those that stayed free. Flag any feature nearly everyone uses as likely table stakes. For each candidate, give the evidence, the risk to free signups, and what data would change your mind.” 

The quick prompt gets you a list of popular features. The better prompt asks for the evidence behind each pick, and for what would change its mind. 

Step 2: Add competitive context 

Next, give the AI your competitors’ pricing pages and ask it to check your candidate gates against them. Here’s how Planloop compared with two fictional rivals: 

 Planloop Taskhaven Stackwork 
Free option Free, up to 10 users Free, up to 15 users 14-day trial only 
Timeline view Free Free Paid only 
Guest access Free Free, 2 guests Paid only 
Lowest paid tier $10 per user/month $9 per user/month $12 per user/month 

This step changed our answer. Timeline view had looked like the obvious gate. But Taskhaven gives it away free, so gating it would have made Planloop look weak next to its closest competitor. 

Guest access was a better candidate. The AI suggested capping free guests at two. That’s exactly what Taskhaven offers, so Planloop would have given up its edge for nothing. I changed the cap to three. The AI didn’t make that call, and it shouldn’t have. It doesn’t know how you want to stand out in your market. 

To pull competitor pricing yourself, try this prompt from the session: 

“Review the public pricing pages for [Competitor A], [Competitor B], and [Competitor C]. Build a table with each tier’s name, price per user per month, user or usage limits, and which of these features it includes: [feature list]. Note where a feature is limited rather than removed. Link the source for every row, and flag anything you couldn’t confirm.” 

Check its sources every time. AI can misread pricing pages, and those pages change often. Save a snapshot of each competitor’s page monthly, or build a competitive research agent that does it on a schedule. Then ask the AI what changed and what strategy the changes suggest. 

Step 3: Design the experiment and size the result 

Finally, ask the AI to design two or three tests for your chosen gate and to estimate what each could deliver. Planloop needed about 18 more upgrades a quarter to reach 16%. The AI estimated the guest cap would deliver about four. One move wasn’t going to close the gap, and that’s useful to know before you commit to anything.

Four SaaS Pricing Experiments That Work in B2B 

Surveys tell you what customers say. Experiments tell you what they do. The bigger the risk of a decision, the more carefully you should pick the test that lowers it. Four SaaS pricing experiments work well in B2B: 

  • New-signup cohort. Apply the change to new accounts only, and leave existing customers as they are. Roger pointed out that you can narrow this further, for example by testing in one geography first. Be careful, though. Existing customers look at your pricing page too, and an unannounced change can make them feel second-rate. 
  • Fake door. Show the upgrade prompt before you build anything, then measure clicks and interest. 
  • Free trial. Start everyone with full features, then move non-payers to the free plan. Measure what they miss. 
  • Deal-desk test. A pilot group of salespeople presents the new package on selected deals, and you track win rate and discounting. It’s private, and it turns that sales team into champions for the change. Be ready to explain what’s in it for them, because any change to a sales process puts their commission at risk. 

How to Design a Pricing Experiment Leadership Will Trust 

A pricing experiment leadership will trust has five parts, each written down before the test starts: 

Part What it answers Planloop example 
Hypothesis If we [change], then [metric] will [move], because [reason] If free accounts get timeline on one project only, free-to-Pro conversion rises, because growing teams hit the limit 
Design Who sees it, for how long, compared with whom? New signups for six weeks, split evenly. Existing accounts unchanged 
Primary metric What shows the gate captures value? Free-to-Pro conversion within 60 days 
Guardrails What must not get worse? Signups, week-two activation and support tickets about limits 
Decision rule What result means ship, adjust or roll back? Ship if conversion improves and guardrails hold. Rethink if activation drops 

Question your decision rule before you set it, then honor it afterwards. If a result goes against you, that’s information too. Don’t explain it away. 

Your experiments may also show that the goal itself is unrealistic. Imagine leadership asks you to lift conversion from 13.5% to 20% this quarter. You map out what each experiment could deliver: maybe a 1% lift from one, and another half a percent from the next. Now you can walk into that conversation with numbers instead of a feeling. Small lifts also need big samples, and proving one can take more signups than a product gets in a quarter. Bring leadership that math too. 

 

Where Pricing Surveys Fit: MaxDiff, Kano, Van Westendorp and Conjoint 

Pricing surveys fill the gap when usage data isn’t enough, and they don’t require you to change anything in the product. They tell you which direction to explore before you run an experiment. 

Method How to run it What it tells you Use it when 
MaxDiff Show sets of four or five features. Each person picks the most and least important Which features matter most You need a ranked list, fast 
Kano For each feature, ask how they’d feel if it were included, then if it weren’t What delights versus what’s expected Separating table stakes from differentiators 
Van Westendorp Ask at what price it’s too cheap, a bargain, getting expensive and too expensive What price range feels acceptable Setting a price for a new tier 
Conjoint Show bundles that mix features and prices. Each person picks the one they’d buy How features and price trade off High-stakes packaging decisions 

Some of these are easier to run than others. Roger has run Kano surveys on his own. For conjoint and the Van Westendorp Price Sensitivity Meter, he’d still bring in a market research expert. A PM needs to know enough to choose the right tool, including the difference between MaxDiff and conjoint. If your company doesn’t have a researcher, AI can take you part of the way. 

 

Beyond Pricing Pages: Five More Competitive Signals 

Your competitors can read your pricing page as easily as you read theirs. These five sources tell you more: 

  • Review sites show what customers resent paying for and why they switched. Expect mostly the happiest and the angriest users, though. 
  • Release notes reveal new features and which tier they launch in. 
  • Help docs often spell out usage limits and plan rules that the pricing page leaves vague. 
  • Job postings for pricing or monetization roles can signal a packaging change. Large competitors hire constantly, so treat this as a weak signal. 
  • Your own sales team hears competitor quotes, discounts and lost-deal reasons every week. If your sales calls are recorded, AI can pull the objections out of the transcripts for you. 

 

Prompting Habits That Make AI a Better Pricing Partner 

AI is an amplifier. It spots patterns across thousands of accounts, summarizes pricing pages and argues against your proposal. Judging what customers will really pay is still your job. So is owning the recommendation. 

These six habits get better output: 

  1. Lead with the goal. Be specific about what you want to move, such as growing upgrades from 13.5% to 16%. 
  2. Say who you are. An early version of our Planloop prompt kept suggesting sales commission changes. Adding “I’m on the product team” kept the answers on packaging. 
  3. Fence the evidence. Say “using ONLY this data”, and ask it to flag what the data can’t support. 
  4. Ask it to push back. AI defaults to agreeing with you. Ask for the strongest argument against your idea. 
  5. Make it show its work. Ask for numbers, sources and a confidence level you define. 
  6. Iterate on the prompt. When the output misses, fix the prompt, not just the answer. Building the Planloop case study took us several tries. 

One more tip from Roger: if your prompt needs an input, add a line telling the AI to ask for it when it’s missing. And only use AI tools your company has approved, especially with customer data. 

 

Why Product Managers Belong in Packaging Decisions 

Product managers bring the context pricing decisions need most. You know why customers upgrade or churn, and which features they lean on. You’re also comfortable turning an uncertain idea into a hypothesis and a test. Finance tends to work in a world where A plus B equals C. Pricing rarely works that way. 

If you don’t own pricing today, you can still earn a seat at the table: 

  • Bring evidence, not opinions. Share a usage analysis with whoever owns pricing. 
  • Write a pricing hypothesis. Keep it to one page: what you’d change, why, and what you expect to happen. 
  • Offer to run the test. Propose a small, low-risk experiment and own the readout. 

You’ll lead the test, but you won’t run it alone. “Product management is a team sport,” as Roger likes to say. You’ll need help from operations or engineering to make sure the test runs properly. For a wider look at where PMs fit, read our piece on the product manager’s role in product pricing. 

 

Your Three Moves This Week 

  1. Pull usage data for your free and paid accounts. 
  2. Run competitive research on your rivals’ pricing pages. If you don’t have an agent tracking them yet, this is the week to build one. 
  3. Bring one gate hypothesis, or one pricing change, to whoever owns pricing. 

 

Watch the Full Workshop for Your Feature Gating Strategy

About The Author

Rina Alexin

Rina Alexin, the CEO of Productside holds a BA with honors from Amherst College and an MBA from Harvard Business School. She is also a member of the AIPMM.

Frequently Asked Questions

Feature gating means restricting a feature in a lower pricing tier, either by removing access entirely or by capping usage. Productside CEO Rina Alexin recommends limited access over a full lockout for many features. A capped feature still lets free users reach value, and it acts like an activation signal that predicts upgrades.
Gate a feature only when three things hold. Larger, growing or paying accounts use it more. Free accounts that upgraded adopted it more than accounts that stayed free. And the market won’t punish you, meaning competitors don’t already give it away free. The most-used feature is usually core, so leave it ungated.
No. Usage is not willingness to pay. A product’s most-used feature is usually a core requirement, like video in Zoom, so gating it can backfire. Look instead for features whose usage grows with account size or separates upgraders from free-forever accounts. Then check competitor packaging before deciding.
Use AI in three steps. First, upload account usage data and ask it to find features that separate upgraders from free accounts. Second, add competitor pricing pages so it can flag moves that would look weak. Third, ask it to design low-risk experiments and estimate their impact. The final call stays with you.
Four tests work well in B2B. A new-signup cohort sees the change while existing customers don’t. A deal desk test is run by a pilot sales team with selected accounts. A fake door measures clicks and interest before anything is built. A free trial measures usage, adoption and conversion directly.

You May Also Like