“I've had to be very budget conscious and time efficient where I am, that means Fivetran first but if that's too expensive for a particular source then we try stitch. If all else fails or if it's a less common source then we go to dlt/dagster.”
“if thats your sources i would recommend Airbyte - its rather cheap (compared to Fivetran or Matillion) and is quite usability friendly (compared to meltano)”
“dlt + Dagster will rein things in a lot in terms of costs and developer experience. Here’s a video showing how I use them for a personal project, basically illustrating an example where you could be migrating from Fivetran/Airbyte or custom ETL: https://youtu.be/QOPy-h4wOl0”
“Fivetran is a great product, but their main problem as with many other tools is a pricing bug. Pricing by row or volume isn't linked to business benefit.”
“I heard Postman switched from Fivetran to Hevo, and it worked out great for them. They saved around 30-40 hours of developer time each month, and even their non-technical team members could set up data sources with Hevo's no-code interface.”
“We've been seeing many customers switch over from Fivetran, with an average cost savings of around 50%. In addition to the savings, they often find more control over their data pipelines.”
“Fivetran was great for our "get up and going" and has served us well, but yes... their pricing model stings sometimes, especially when doing full backfills on production tables with new columns.”
“i would look into [stitch](https://www.stitchdata.com/) i’ve used it before to as an alternative to fivetran.. it’s slightly less expensive, but completely depends on the use case.”
“Opaque data loading process: This is fine for smaller jobs, but I have had a table with gigabytes of data, and the load part ran for seven days without visibility of how long it would take. Their customer support didn't know either.”
“If that's the case we will have to give our entire operating margin (>10 billion ) to Fivetran. It's probably that much for a million rows or something”
“As a small company, Fivetran is a huge time saver in allowing us to get data where it needs to go fast and 98% of the time I love it. The other 2% I'm wading into a sea of failures, warnings, and alerts that don't provide enough detail.”
“Because of those dependencies, the sync failed, the tables look like hell with deprecated columns scattered about, PK indexes on those columns dropped, and queries performing like hot garbage.”
“Fivetran charges by "monthly active rows", which quickly adds up when you have hundreds of millions to billions of rows that are constantly changing.https://fivetran.com/docs/usage-based-pricing”
“We use fivetran, dbt and snowflake and I don't really see a reason to have an orchestration tool. Fivetran and dbt do a great job of orchestration on their own in my opinion.”
“Fivetran can integrate with DBT Core directly to schedule model runs after the connector refreshes. Managed it as a team of one at my last startup to Snowflake - pricey but worthwhile for us due to the ease”
“Just load your data using Fivetran into their S3 destination connector that adds an iceberg/delta metadata layer on top. It’s pretty performant and pretty cheap. They’ll manage that raw layer and then use whatever service you want on top”
“Snowflake also have a connector to GA4 (aggregated, not raw) which will use the same GA4 APIs as Fivetran. This may be worth a try as well in case it is a quirk of Fivetran causing your issues.”
“From what I've seen, there's no page title available at all within any of the tables from Fivetran's connector, and I can't get any of the data to align with what we see in GA4.”
“The particular pull for Airbyte is that we can self-host the open source version, rather than pay Fivetran a significant sum to do this for us.It's an immature tool, so I don't yet know that I can claim we've spent _less_ than Fivetran on the additional engineering and ops time, but it feels like it has potential to do so once stabilized.”