Frequently asked questions

Metricsense reads all of your customer feedback, whether it arrives as app store reviews, support tickets, survey responses, call transcripts or a CSV, and answers questions you write in plain English. Every answer links back to the customer's own words, so you can check it. On feedback you are reading insights in minutes.

You are not choosing from a list of categories somebody else wrote. You ask the question: "What are paying users complaining about since the last release?" It runs across everything you have uploaded and comes back with the answer and the quotes underneath it. No tagging, no taxonomy to agree on first, no model training.

Product managers who need signal out of app reviews. CX and support leads who already know what the problems are and cannot yet prove them. Founders and CPOs deciding what to build next on something firmer than the loudest recent complaint.

App Store reviews, Google Play reviews, call transcripts from AWS Connect, NPS and CSAT responses, and anything at all as a CSV. Any source with an API can be connected.

Three ways. Connectors for the App Store and Google Play that sync on their own. A feed from a system like AWS Connect for call transcripts. CSV upload for everything else, including exports from a helpdesk or survey tool you already run.

As far back as the source will give you. For the App Store and Play Store that is whatever history the store still holds. For a CSV there is no limit: upload everything you have kept.

Every answer opens onto the customer quotes underneath it. If Metricsense tells you checkout crashes are your top complaint, you can open that theme and read every review that put it there. Nothing is a number you have to take on faith.

Yes. It tracks how sentiment and themes move, and alerts you by Slack or email when something spikes. Ask "how has sentiment about checkout changed this quarter" and you get the trend and the feedback behind each point on it.

Yes. Metricsense writes Jira tickets and GitHub Issues from the evidence. Each one carries the customer quotes and the counts that justify it, so the engineer reading it can see why it exists, not just what it is called.

Whatever you can describe. "Classify the topic: billing, onboarding, performance or UX." "Rate the urgency: low, medium, high, critical." "Is the customer naming a competitor?" Each one becomes a dimension you can chart, filter and set an alert on, applied the same way to every item.

Yes, and it scores both sides. The customer side: complaints, frustration, the moment a call turns into an escalation. The agent side: whether the process was followed, how the customer was handled, where the resolution went wrong and what to coach.

There is a free tier with 500 analysis credits and no credit card. Paid feedback plans are Starter at $19 a month and Growth at $59 a month, with more credits, alerting and export. Enterprise is on custom terms.

Call analysis is not on the fixed plans. It is priced per customer on your call volume, because it is deployed into your own cloud and the concepts are built from your quality process. The pilot that comes first is free.

For feedback, minutes: upload a CSV or connect your store listing and start defining insights. No tagging, no taxonomy, no training period. Call analysis is a different shape of work, and starts with a one-week pilot on calls you already hold.

No. You describe what you are looking for the way you would explain it to a colleague. "What are users complaining about since the last update?" No SQL, no prompt engineering, no data scientist on standby.

Metricsense lets you define any insight and returns answers linked to the raw customer quotes, and it scores both the customer side and the agent side of call transcripts.

Yes. Upload the CSV you already keep and start asking questions of it, with no tagging and no taxonomy to agree on first. Nothing has to be restructured before it will run.

For structured, repeatable analysis, yes. Metricsense applies the same insight definition to every item, tracks it over time, and links each answer to the original quote. ChatGPT is better for a quick one-off look at a small set.

Your feedback is not shared with AI model providers for training. You control retention and deletion. Avesta Labs, which builds and runs Metricsense, holds a SOC 2 Type I attestation, and the Type II observation period is under way.

In your own sovereign cloud, in your tenancy and your region. Your recordings do not move; the software goes to them. The model runs on Amazon Bedrock inside that same boundary, and PII is stripped before any transcript reaches it.

What is collected, where it is stored, how long it is kept and how to have it deleted are all set out in the Privacy Policy, linked below and in the footer of every page.