When a subscription performs differently across partners, check five parts of each bundle: who the partner reaches, where the offer appears, the offer itself, how people activate and how they pay. A fair comparison shows where results differ, and a controlled test can help establish what caused the difference. Then test what extra investment adds before moving budget.
The same subscription can perform differently from one partner to the next
Your service is live with a dozen partners across several countries. One partner’s subscribers activate and keep watching, while another’s sign up but fade away. Leadership wants to know which partnerships have earned more investment.
Answering satisfactorily requires you to separate what you can see in the results from what you can show caused them.
Which parts of a bundle should you check when results differ?
There are five: the audience, the placement, the offer, activation and billing.
- Audience. Who the partner reaches. Are they the people most likely to value your service?
- Placement and timing. Where and when people see the offer. Does it reach them at a moment when they’re ready to decide?
- The offer. Its tier, price and any free period. Does it give people a reason to start, and a reason to stay?
- Activation. Whether people who accept the offer go on to use your service. Are they getting started, or signing up and stopping there?
- Billing. How subscribers pay, which in a bundle runs through the partner’s billing. Is it easy for people to keep paying?
Results can differ for reasons outside the bundle, such as how well people in a market know your brand, local competition, the season or how actively a partner promotes the offer. So, treat the five parts as places to look, knowing more than one may matter. Our blog on audience quality vs reach explains why a partner’s biggest audience isn’t always its best one.
Compare partner performance, then test why it differs
Partners vary in market, audience and offer, so raw totals can mislead. Evaluate every partner the same way and at the same point after launch, using the measures set out in our blog on how each category fits partner offers.
Comparisons can also overstate what worked. In a study of 663 advertising campaigns published in Marketing Science in 2023, even advanced statistical methods without a randomly chosen comparison group produced much larger estimates of impact than randomized tests.
The clearest evidence comes from a simple test. Agree with a partner to hold back a random group of eligible customers who won’t see the change, then compare their results with everyone else’s. Even then, the result only applies to that partner and market, and may not carry over to others.
When a controlled test isn’t possible, change one thing and treat the result as a guide
Not every partner can hold customers back. Where this is the case, compare it with an earlier offer, a different launch period or a similar market, which can still guide the next decision, although other things will have changed too.
This is where subscription optimization in bundles differs from testing on your own website. Each launch is a commitment, and live offers stay as agreed with the partner. So change one thing at a time. You’ll see what made the difference, though one launch won’t tell you everything.
Offer design affects trial-to-paid conversion, so test it with each partner’s customers
The length of a free period shows why offer design is worth testing. In a randomized study of 337,724 users of a software service, published in Management Science, a 7-day trial brought in 5.6% more subscribers than a 30-day trial. In a 2025 experiment with 680,588 users of an image-editing service, a 7-day trial converted more users than a 3-day one, and the effect varied with users’ cultural and economic context.
Neither study looked at partner bundles, so neither sets the right free period for your partners. What they show is that offer design can move conversion, which makes free periods worth testing with each partner’s customers, and price testing worth considering too.
A strong average doesn’t show what extra money would add
When leadership asks where to invest moreput more money, the partnership with the best results so far isn’t necessarily the best place for the next dollar. Start with the subscribers each partnership brings who activate and stay, and what they earn after the partnership’s costs. Our guide to using data to manage bundling partnerships covers the reporting behind that view.
Then test the extra investment itself. Increase it in one place, such as more promotion with one partner or a richer offer in one market, and compare the extra subscribers who stay with a group that didn’t get the increase.
Boku’s reporting gives subscription businesses figures at market, partner and offer level across the partners they work with through Boku, so each comparison uses the same measures.
Subscription bundle optimization means investing where extra money adds the most
Back to those twelve partnerships. Their results so far show where to investigate, and the decision about more investment depends on what extra spending returns. Test that return in one place before committing more budget.
FAQs
What is subscription bundle optimization?
It’s the work of improving how bundles perform across partners. That means comparing results on equal terms to see where they differ, testing changes in later offers to learn what made a difference, and putting more money where extra investment adds the most subscribers who stay.
What makes one subscription bundle perform better than another?
Five parts of a bundle are worth checking: who the partner reaches, where and when people see the offer, the offer itself, how people activate it and how they pay. Factors outside the bundle, such as brand awareness, competition and the partner’s promotion, can matter too.
How do you measure bundle performance across partners?
Use the same measures at the same point after launch for every partner, and add what each partnership earns over time after its costs. That shows where results differ. To estimate what caused a difference, compare against a randomly chosen group that didn’t get the change, where the partner agrees.
What is a good trial to paid conversion rate?
There’s no universal benchmark, because it depends on the service, the partner, the market and the offer. Compare each offer with your own earlier offers and with your other partners on the same measures. Research shows the design of a free period alone can change how many people convert.
How do you decide which partnerships deserve more investment?
Look at how many subscribers each partnership brings who activate and stay, and what they earn after costs. Then try the extra investment in one place and compare it with a group that didn’t get it. Strong results so far don’t show what additional money would add.
Key takeaways
- When results differ across partners, check five parts of each bundle: audience, placement, offer, activation and payment, plus factors outside the bundle.
- A fair comparison shows where results differ. A controlled test, with a randomly chosen comparison group, can help establish what caused a difference.
- Where a controlled test isn’t possible, change one thing in the next offer and treat the result as a guide.
- Before moving budget, test what extra investment adds, since strong results so far don’t show the return on more money.
