What do you dislike about Apify?
Our experience with Apify’s Twitter scraper started positively, but over time it became extremely disappointing and costly due to multiple issues that directly affected reliability and budget control.
1. Parameters not working as expected
Many of the parameters provided by the actor did not behave consistently. Even when we configured limits such as maximum items to fetch, the scraper did not always respect them. This made it very difficult to rely on the actor in a production workflow where predictable behavior is critical.
2. Unnecessary fetching that wasted our budget
On several occasions the scraper fetched thousands of tweets even though strict limits were configured. These runs consumed a large amount of resources and unexpectedly increased our costs. What made this worse was that these fetches were not intentionally triggered by us, yet the platform still charged for them. When we raised the issue, there was no meaningful resolution or refund, even though the behavior clearly went beyond the configured limits.
3. Fetching outdated tweets instead of recent ones
Another recurring issue was that the scraper frequently returned old tweets instead of the latest ones, even when using options intended to retrieve the most recent results. For time-sensitive workflows this makes the data unreliable. We often saw situations where tweets from the previous day appeared while tweets posted within the last hour were missing entirely.
4. Uncertain and inconsistent scraper behavior
The overall behavior of the Twitter scraper felt unpredictable. Identical configurations would sometimes produce completely different results between runs. Some runs would miss relevant tweets, while others would return irrelevant or outdated data. This level of inconsistency makes it difficult to trust the tool for automated systems.
While Apify provides a capable platform and a developer-friendly interface, the lack of strict control over limits, unreliable scraping results, and poor cost safeguards created serious operational issues for us. For any system that depends on predictable data collection and controlled spending, these problems can become very costly very quickly.
Until stronger safeguards, clearer parameter behavior, and better cost protection mechanisms are implemented, it is difficult to recommend relying on the Twitter scraper for production-critical workflows. Review collected by and hosted on G2.com.
