Forecasting how many conversations with your users will AI Copilot will fully resolve – and ultimately estimating what what will be total monthly Copilot cost – can feel uncertain or fuzzy at first. This article provides help and guidance to estimate copilot resolutions – whether you already use some sort of support chat or helpdesk (most likely with human agents), or whether Copilot will be your first support conversation solution.
First, two quick definitions (plain English):
#1 If you already have support history (easiest) – use what you know from chat/helpdesk:
#2 No history? Use MAUs and a simple contact rate – when you don’t have past tickets, estimate them from usage.
Keep it safe & real
If you are new to Copilot and Resolutions topic, we recommend to scroll below and read the entire article. If you just want to use the Copilot Resolutions Estimator, go ahead.
Quick guide: This mini-calculator turns what you already know about your support volume into a simple monthly forecast. If you are not sure about the support volume for your app or SaaS, read below tips how to estimated it.
What you enter
What you get
Historic tickets × ACR (how many conversations Copilot will close).Overage × price per Resolution. – What extra you will pay monthly above your plan (note that Overages are billed monthly even for annual plans).Copilot is an AI assistant capable of resolving a high percentage of support questions automatically before they reach your team. Think of it as a support teammate embedded in your product. It answers “how do I…?” questions using your existing help content (knowledge base, FAQs, release notes). It doesn’t replace your team; it handles the repetitive, well-documented questions so humans can focus on higher-value work.
A Resolution is a fully solved support conversation—the same way you consider a helpdesk ticket “closed.” In Product Fruits, a conversation is counted as resolved when the exchange ends positively and the user doesn’t request a human or leave negative feedback within a short window (e.g., 24 hours). You can read more about it in a dedicated article.
All current Product Fruits price plans include a certain number of free copilot resolutions (20 or 40) – and you can negotiate a custom included volume in the Enterprise plan. Resolutions above this quota are billed at $0.69 per resolution (or at the negotiated / custom rate for Enterprise plans). There are unlimited* resolutions during the Copilot trial period – this applies both for existing and new customers. This unlimited resolutions policy allows you to fully roll out and test Copilot with all your users – and allows you to better estimate actual resolutions and costs in the later production stage, especially if you have no historic data for user support conversations or tickets.
*There is actually a fair usage quota of 3000 resolutions; we believe this makes the trial unlimited, for virtually all our customers.
The idea is to convert familiar numbers – your historical monthly tickets, contact rate or support conversation – into an estimate of Copilot usage and in particular user support conversations resolved by Copilot itself (without human intervention) – Resolutions. This then translates to a monthly cost (and possibly savings) estimate.
Think of Copilot as first line of user support in chat channel. So tickets or conversations that originated from in-app chat will go through Copilot first. The share of those monthly tickets or conversations your Copilot can close end-to-end once it’s connected to your knowledge base is your resolutions estimate.
The actual percentage of course varies depending on quality of knowledge base and other content used by Copilot and may also change (ramp up) over time as users learn to use Copilot as their main source of support and self-help. This would most likely lead to a reduction of support requests through other channels like e-mail or voice, and significant potential savings.
At its core, the resolutions estimate calculation uses one line of math:
Monthly Resolutions = Historic Tickets (Chat) × ACR
For Historic tickets Start with the last 2–3 months’ average, to set starting ACR, we recommend measuring copilot performance during second week of free testing. You may start around 30 % or even less, but even with this starting point, expect to ramp up to 40-60% after a month or two – especially if you have good quality support content such as knowledge base.
Think about the overall Copilot resolutions estimate is a range, not an exact number. To obtain a final cost range estimate, you should consider your plan and resolution price – either the standard $0.69 or a lower negotiated rate for the Enterprise plan.
If you don’t have past support data, no problem. Start with two simple things:
Pick a contact-rate range that fits your type of app (we share handy bands just below), then choose a starting ACR—that’s the Automation Capture Rate, or the % of questions Copilot can fully solve on its own (without handing user support request over to human support chat or email). Put those together and you’ll get a ballpark number of monthly Copilot Resolutions and the expected cost.
Finally, prove it in our free two-week Copilot test (with unlimited resolutions). Week 1 is usually setup and launch, so use Week 2 to see real early usage and adjust your numbers accordingly. You can then use the Resolutions Estimator above to get an estimated monthly cost.
| App type (indicative) | Typical contact rate (tickets ÷ MAUs, monthly) | Internal notes |
|---|---|---|
| Consumer / mobile utilities | 0.3–1.0% | High self-serve; release spikes. |
| PLG B2B SaaS (self-serve SMB) | 1.0–1.8% | Onboarding & billing dominate. |
| Complex B2B SaaS (mid/enterprise) | 1.5–3.5% | Deeper workflows; integrations. |
| Dev tools / APIs | 0.5–1.5% | Lower average, but bursty on launches/migrations. |
*Why “indicative”? Public sources confirm the metric itself (contact rate/tickets per user) but don’t publish stable cross-industry MAU ratios—so use the ranges above to size the ballpark, then adjust with your own data during the free Copilot trial/test period. (The table above was compiled from various sources, including Gorgias, MetricNet, ThinkHDI, and Zendesk.)
Note: If you want to be extra-conservative, you can use the guideline of SaaS support tickets ceiling at <10% of active users as referenced by 8020consulting (consider it explicitly as a ceiling, not a norm).
Two simple tools keep spend predictable: the “Maximum Resolutions” cap and your two-week trial data.
Set a monthly ceiling for billable Resolutions. When you hit it, Copilot pauses automatically and stays hidden until the next billing cycle—or until you raise the cap. Even with Copilot hidden, your users will still have access to AI-powered Knowledge base search and to human support chat (if enabled). You’ll also get an in-app alert at 80% of your cap so you can adjust early.
Quick starting rule: Suggested cap = your expected monthly Resolutions + 15–25% buffer.
Example: expect ~200 → set cap 250. You’ll see an alert at 200 and can decide whether to raise the cap or let Copilot pause.
As first couple of days may include setting up and testing Copilot by your own team, focus on trial days when Copilot was fully available to your users.
*In case user had no more questions, did not ask for transfer to human and did not show negative sentiment.
Practical tips
With a reasonable estimate, hard cap, an early alert, and one realistic week of data, you’ll have tight control over spend—without sacrificing the value of always-on AI support.
