返回资讯中心
外部精选
软件工程
#开发者实时流

Show HN: Overhear – mention monitoring billed per search, not per month

X, Reddit, LinkedIn, YouTube and the web, checked at the pace you set. Complaints, questions and praise sorted for you, plus the ones asking whether to buy.

Hacker News 最新hauju2026/11/1 · 今天4 分钟阅读

以下正文同步自 Hacker News 最新,版权归原站所有,已转换为易读排版。

X, Reddit, LinkedIn, YouTube and the web, checked at the pace you set. Complaints, questions and praise sorted for you, plus the ones asking whether to buy.

1,000 free credits to start, no card needed. Then prepaid credits, no subscription.

Cheaper alternative to acme for a team of three?

X, Reddit, LinkedIn, YouTube, Hacker News and the open web, searched in parallel. X and Reddit every hour, LinkedIn every six, YouTube, Hacker News and the web daily. Can't wait? Scan now

We scanned our own product, SeggWat, once. Every source reports what it found, what it cost in credits and how long it took.

Overhear gives every mention one label, so the posts worth answering don't sink under the rest.

Our team of three is outgrowing spreadsheets. Is acme worth it, or is there something cheaper that does the same?

Four days and still no reply from acme support about a double charge.

The acme CSV export stops after exactly 1,000 rows. Anyone else?

Does acme have an API, or is everything in the UI?

Moved our onboarding to acme last week. Setup took one afternoon.

Three brand keywords at the default cadence: ≈ 7,920 credits a month, ≈ $13.20 at the $10 pack rate, with no subscription. Stalkr lists Starter at $29 a month for 3 keywords, Pro at $99 and Scale at $299. See pricing

1,000 free credits to start, no card needed.

Every search is one metered call with a receipt, so you see what each scan cost.

正文由 探How 从来源站点 RSS 同步,内容未经改写;遇到排版缺失或需要图片、视频时请以原文为准。

打开原文
去社区讨论 →