Competitor and product review explorer for release planning

I want an app my product team opens when we are planning a release. It is a review explorer that covers both our own products and our competitors' products, so we can judge how we are doing relative to them instead of only looking at ourselves.

The main screen has two controls: a product picker where I select one or more products, ours and competitors' together, and a date range picker. Everything else on the screen respects those two controls.

With products and a date range selected, show three things. A rating trend chart plotting average star rating over time with one line per selected product. A review volume chart showing how many reviews landed in each period per product. And a ranked list of the recurring themes in those reviews.

The theme list is the centre of the app. Group the reviews in the selected range into recurring themes, for example battery life, shipping delays, customer service, price. Rank the themes by how many reviews mention them, and give each theme a direction label of rising, fading or steady. Work the direction out by comparing the share of reviews mentioning that theme in the more recent half of the selected range against the earlier half. Break the counts out per product, so a theme that is hurting a competitor but not us, or the reverse, is obvious at a glance.

Clicking a theme opens the verbatim quotes behind it. Each quote shows the exact review text, the star rating, the review date and which product it came from. This is the whole point of the app: nobody should be able to argue about whether a complaint is real, because the underlying reviews are sitting right there. Let me filter these quotes by star rating and by source, and let me select individual quotes.

Bright Data supplies all of the review data. When I add a product, use Bright Data's List Marketplace Datasets to discover which review datasets the connected account actually has access to, and let me choose the right review source from that list rather than hardcoding dataset ids. To pull an Amazon or Trustpilot review set, use Trigger Dataset Collection to start the job, poll Check Snapshot Status until it reports ready, then use Download Snapshot Results to fetch the reviews. For any other review site, let me paste in the address of a review page and fetch it with Scrape URL with Web Unlocker, then parse the reviews out of the page that comes back.

Bright Data only keeps snapshots for about 16 days, so once reviews are downloaded, store them in the app's own storage keyed by product, source and collection date. Never rely on re-reading an old snapshot id. Give me a refresh control per product that runs a fresh collection and merges the results into the stored set without creating duplicates, and show when each product's reviews were last refreshed.

Add an Export evidence pack button. It takes the quotes I have selected plus the theme counts for the current selection, and appends them to a Google Sheets spreadsheet using Append Values, one row per quote carrying the theme, the product, the star rating, the date and the review text. It then posts a short summary to our product channel with Slack Send a Message: the products and date range compared, the top themes with their counts and direction, and a link to the spreadsheet.

This is an exploration and analysis surface, not an alerting one. Favour filters, side by side competitor comparison and drill down into quotes. Do not build notifications, a per review triage queue, scheduled runs, or anything that routes individual reviews to owners. Someone opens this app when they need to build an argument for a release plan, and every part of it should serve that.