In this article

An add-on can sell well and still leave a software business earning less. Some customers buy something they previously could not justify. Others use the new option to replace a more expensive plan. The same purchase count includes both stories.

Before separating a feature from a subscription tier, model the customer’s previous bill and the cost of serving the new package. The useful question is how much contribution the packaging change creates across the existing customer base, after substitution and delivery costs. A strong attach rate by itself cannot answer it.

Start with the customer job you are separating

An add-on makes the most sense when it serves a recognizable need that varies independently of the core plan. A team might need more advanced outbound messaging without needing every administrative capability in a higher tier. Another team might need advanced governance and never use that messaging feature.

Intercom’s Proactive Support Plus is an example of that structure: its documentation describes an optional package available alongside its core plans, with a base fee and usage charges for some capabilities. That demonstrates a packaging pattern, not proof that the same package would work for another business.

Write a one-sentence purchase reason before opening a spreadsheet. “Teams buy this when they need to run recurring customer education” is more useful than “We need another revenue stream.” It gives research a specific situation to investigate and tells product which capabilities must work together.

Then name the people who already receive the feature. If it is the main reason customers pay for the higher tier, offering it separately creates an obvious route down the pricing ladder. That route may still be worthwhile, but its cost belongs in the model.

A monthly model that includes the downgrade

Consider a fictional product with 800 Base accounts paying $60 a month and 200 Plus accounts paying $100. Plus includes the feature under consideration. The proposed add-on costs $15 a month and would be available on Base; Plus keeps it included.

For this illustrative steady month, assume:

  • 120 existing Base accounts buy the add-on.
  • 30 existing Plus accounts move to Base plus the add-on.
  • Each new Base adopter creates $5 of additional monthly delivery and support cost.
  • The add-on creates $300 of additional shared monthly support and administration cost.

The downgraders already use the feature, so this model assumes their delivery cost is unchanged. It also assumes no customer acquisition, cancellation or other plan change. These are inputs to test, not benchmarks.

Monthly changeCalculationContribution change
New purchases from existing Base accounts120 × $15+$1,800
Additional cost of serving those adopters120 × $5−$600
Revenue lost from Plus downgrades30 × ($100 − $60 − $15)−$750
Additional shared operating costStated assumption−$300
Net change$1,800 − $600 − $750 − $300+$150

A report showing $2,250 in add-on revenue would include all 150 buyers. It would miss the $1,200 of Plus-plan revenue displaced by the 30 downgrades, as well as the additional costs. The model’s $150 contribution improvement is the relevant result under these assumptions.

This is an incremental operating comparison. It does not include development investment, taxes, payment fees, changes in collection or the business’s full overhead. Add those items when they change because of the proposal; keep their treatment consistent in both scenarios.

Solve for the attach rate you actually need

Every new Base adopter contributes $10 after the assumed $5 variable cost. The expected downgrade loss and shared operating cost total $1,050. Break-even therefore requires:

($750 + $300) ÷ ($15 − $5) = 105 new Base adopters.

That is 13.125% of the 800 eligible Base accounts. The assumed 120 adopters represent 15%, leaving only 15 accounts of headroom above break-even. A small error in the demand estimate could eliminate the improvement.

Downgrade behavior matters just as much. If 40 Plus accounts move down instead of 30, the loss becomes $1,000. With 120 new Base adopters, contribution becomes −$100 after the same costs. More customers buying the add-on would coexist with a worse economic result.

Use a small sensitivity grid before discussing the forecast as a single number. Vary new adoption, downgrade count and cost per adopter. Keep the original assumptions visible. If the proposal works only at the most favorable intersection, the next step is a narrower test or a different package.

Look for the substitution in customer research

Ask current higher-tier customers what they would remove if the feature became available separately. Show the whole proposed offer, including the capabilities they would lose by downgrading. Asking whether they like a $15 option without showing the tradeoff produces a weak signal.

For Base customers, investigate the last occasion when the missing feature prevented a useful task. Identify whether they solved the problem elsewhere, deferred the work or simply did not need it. Interest in a cheaper feature and a willingness to change an operating workflow are different commitments.

A test should distinguish existing Base adopters, higher-tier downgraders and genuinely new accounts. Record the prior package before the offer is shown. Compare contribution over a complete billing interval, then follow renewal. An initial purchase that is removed before the next cycle has a different value from sustained use.

Where the team can run a controlled offer test, assign eligibility in a way that avoids selective exposure to the easiest buyers. If the test is a limited commercial rollout instead, describe it that way and retain the selection criteria. Neither approach makes stated preference equivalent to collected revenue.

Make the package operable before expanding it

The feature needs an access rule that works when customers add it, remove it or move between tiers. Stripe’s entitlements documentation describes mapping product features to customer access. Your product still needs to decide what happens to work created with a feature after access changes.

For example, a customer might retain existing reports while losing the ability to create new scheduled exports. A higher-tier plan that already includes the feature should not accidentally charge the customer for it twice. Support should be able to explain the current package without reconstructing a sequence of manual exceptions.

Choose the pricing metric separately from the packaging boundary. A feature can be a useful add-on but require a usage allowance or another charging unit to remain sustainable for heavy users.

Finally, give the proposal a decision rule. Under the example’s assumptions, the add-on needs more than 105 new Base adopters to create monthly contribution. If observed downgrades rise, recalculate that threshold. The point of the model is to make it easy to change the decision when the customer mix changes.

Sources

Something we should clarify?

Suggest a correction