Datasets and daily data

How Ivy Link collects your site's data every day, how it reaches your repo, and how to check that it's fresh and live.

What a dataset is

A dataset is your own data, collected for you. Ivy Link collects it from public sources every day and stores each day's data, so its history builds up over time. Each item in a dataset is a reference: one thing your pages are about, such as a product model, with its stats.

Each day, Ivy Link writes the latest data as files in the data/ folder of your site's GitHub repo. Your pages are built from those files. They show the data as tables, charts and written answers.

A few rules keep your pages accurate:

Answers come from the data. The numbers and answers on a page are computed from your dataset, not written by hand or guessed. If the data can't answer a question, the question stays off the page. Rankings only use figures with enough sales behind them.
Data only grows. A failed or thin day adds nothing and removes nothing. If a reference gets no new data one day, its page keeps the data and date it already had.
Dates are honest. Each daily-updated page shows Updated with a date. That's the day its data last changed, not the day collection ran. A day that changes nothing keeps the old date. Your sitemap uses the same date.
Pages with data stay in your sitemap. A quiet week or a failed collection doesn't drop a page that has data from your sitemap.
No raw downloads. Pages never link to raw data files.

If your site has no dataset, none of this shows in your dashboard. The Data updates setting only appears once your site has one. (On accounts set up before the current sign-up flow, Home's older overview also has a Datasets section.)

How data is collected and refreshed

Collection runs every day. You don't need to start it.

Ivy Link collects the day's data for each reference and stores it.
It publishes the run to your repo, the way your Data updates setting says: automatically, through a pull request, or not at all while paused.
Your site rebuilds with the new files, and the new numbers are live.

If publishing fails, the data is still stored. Ivy Link retries every five minutes, using the latest stored run.

The daily checks that your dataset updated and that your pages are live are free. They don't use credits. See Credits.

Data updates in Permissions

Data updates decides how the daily data files reach your repo. It's on Settings > Connections, under Permissions. The hint reads The daily data files in data/ that your pages are built from.

Option	What happens
Automatic (the default)	Ivy Link commits the day's data to your repo each day and your site rebuilds with it. It only changes files in `data/`.
Pull request	Ivy Link keeps one pull request with the latest data, updated each day. Merge it when you want the new numbers live.
Paused	Ivy Link keeps collecting but doesn't write to your repo. Turn it back on and the next update brings everything in.

To change it:

Open Settings > Connections and scroll to Permissions.
Under Data updates, pick Automatic, Pull request or Paused.
Click Save data updates.

You'll see Saved. Data updates: automatic. (or pull request or paused). The change applies from the next data run.

Only the account owner can change this setting. See Settings.

Pausing data updates doesn't stop collection. Nothing is lost. When you turn updates back on, the next update brings in everything collected while you were paused.

The standing data pull request

With Data updates set to Pull request, Ivy Link keeps one open pull request per dataset in your repo. It's titled Data update: followed by the dataset name and through the date of the latest run.

Only files in data/ change.
It stays open and is updated each day with the latest data.
Merging it makes those numbers live. Your Live on your site date moves to that run.
If you close it without merging, a new one opens on the next run.

To publish newer data:

Find the data pull request on Decisions (Review the collected data update before it is merged.) or on GitHub. On older accounts, Home's overview also shows Newer data is waiting in a pull request for you to merge. on the dataset's card, and New data is waiting in pull request under Recent work.
Open the pull request on GitHub.
Review it, then merge it.

The pull request's description has a link back to Settings if you'd rather publish automatically.

In Pull request mode, collected data only counts as live once you merge. If the pull request waits three days or more, the daily checks note that newer data is waiting.

Check whether your data is fresh and live

The current Home has no dataset card. To see freshness, check the Data collection section of your weekly report (see below). On accounts set up before the current sign-up flow, Home shows an older overview with a Datasets section. There, each dataset has a card with its name and these lines:

Line	What it means
N references · N days collected · updated daily	How many references your dataset holds, and how many days of data are stored.
Latest data: date	The newest run Ivy Link collected. It may add with sales for N references, and how many failed that day.
Live on your site: data through date	The run your site shows now. Click commit to see the exact change on GitHub.

What each state on that older card looks like, and what to do:

You see	What it means	What to do
Latest data: waiting for the first run	Your dataset is set up but hasn't collected yet.	Nothing. Check back after the first daily run.
Live on your site: not yet	Data is collected but hasn't reached your site yet.	On Automatic, wait for the next publish. On Pull request, merge the data pull request.
Live on your site: data through date (commit)	Your site shows data up to that date.	Nothing.
Newer data is waiting in a pull request for you to merge.	You're on Pull request mode and a newer run is ready.	Merge the pull request when you want it live.
Data updates are paused, so your site shows data through date	You paused updates. Collection continues.	Click Turn them back on to go to Permissions.
The latest data is stored. Site publication needs attention and will retry.	Publishing to your repo failed. The data is safe.	Nothing at first: Ivy Link retries every five minutes. If it persists, check your GitHub connection.

On a healthy site, Latest data is today or yesterday, and Live on your site is close behind.

What the daily checks flag

Ivy Link's daily checks watch your dataset for you. They skip a dataset while its updates are paused. They flag:

No new data since a date. A dataset should have a run from today or yesterday. It should update every day.
Collected but not live. On Automatic, the site is two or more days behind the latest data, or the data has never been published.
Waiting in a pull request. On Pull request, the site is three or more days behind.
References with no new data. Some references got nothing in the latest run. Their pages keep the data already stored. The check says more when the same references come up empty three runs in a row, or when a reference has no stored data to show.
A page whose Updated date is behind its data. The date on a page is two or more days off from the data it was built from.

The latest checks are on Notes as Daily checks entries. (On older accounts, Home's overview also lists them under Recent work.) The two most serious, no new data and collected but not live, also show on Decisions as A data check needs attention, with the note This check needs review before work is added. Open Approvals to review the next steps.

Approve new data, or a page or tool built on it

Your agents can suggest growing your dataset, or building on it. These ideas wait on Approvals like any other. There are two kinds. On Decisions they show as Suggested work, like any other idea.

Kind	What it is
Data	More data for your dataset, such as more sources or a new field.
New page	A page or tool built from your data, to answer a question your site doesn't answer yet.

To decide:

Open Approvals.
Read the idea's why and its Because line, if it has one.
Click Approve to start it, or Drop to have your agents leave it.

Once you approve an idea, its work shows up on Results.

If New ideas is set to Start without asking, agents start data updates and new pages on their own. You can still pause or drop any of them, and they only spend within your budget. See Approvals.

Data pull requests also appear on the Decisions page with Review the collected data update before it is merged. Under Site changes, a data pull request's Publishing: line reads Your review while it's open, then Merged or Closed.

What you see on an older account's Home

The current Home has no Datasets or Recent work section. It shows Now, Needs you, This week and Done lately. On accounts set up before the current sign-up flow, Home shows an older overview, where data also shows up in Recent work:

Entry	What it means
Updated dataset with N references / Data for date, live on your site (commit).	A run was published and is live.
Collected dataset data for date / It goes live on your site once it’s published.	A run is stored but not on your site yet.
New data is waiting in pull request #n / Merge it when you want the new numbers live.	A data pull request is open for you.

What you see in weekly reports

Each weekly report has a Data collection section. For each dataset it shows:

Field	What it means
Latest received run	The newest run Ivy Link received by the end of the week.
Latest published run	The newest run Ivy Link published to your repo. In Pull request mode, that's the data pull request, which still has to be merged.
References in latest run	How many references the latest run covered.
Failed references	References that got no new data in that run.
References retained in run	References that got no new data but kept the data already stored.
Data through	The date or date range the data covers, or Unknown if no date is stored.

You may see Data dates are unknown for a number of references, when some have no date. Each dataset ends with A received run does not establish that its data is live. Received data still has to be published to show on your site. Compare Latest received run with Latest published run to see if publishing is behind. To see what your site shows now, compare those two lines, or on an older account check Live on your site on Home's overview.

If your site has no dataset, the section says No dataset is connected.

Questions

Does Ivy Link promise my data pages will rank? No. Fresh, accurate data gives pages something worth citing, but no one can guarantee rankings.

Can I stop data from changing my repo? Yes. Set Data updates to Paused. Collection continues, so nothing is lost when you turn it back on.

Does Ivy Link change anything outside data/ with data updates? No. Data updates only change files in data/. Changes to pages go through How finished work reaches your site. See Ship changes.

Questions about your dataset? Email hello@ivyl.ink.

Last updated October 31, 2026

Something here not match what you see? Email hello@ivyl.ink and we'll fix the page.