There is no button that turns a CSV file into hundreds of good videos on its own. A CSV holds the changing parts, such as names, prices, photo links and colors. A video needs a designed template to put them into. The reliable path is one After Effects template rigged with Templater, one row of data per video, and a renderer that works through the rows while you do something else.
That answer disappoints people who searched for a “CSV to video converter” and expected to drop a file onto a web page. It should also relieve them. The part a converter cannot do, the design, is the part you already own. The part left over is plumbing, and plumbing can be planned. This guide is written for the producer, marketer, developer or agency lead holding a spreadsheet with a few hundred rows and a deadline. It covers what each piece does, how a CSV actually gets into Templater or Que, how to build the file and the template so a long batch survives, and when this whole approach is the wrong choice.
What a CSV can and cannot do for a video
A CSV is a plain text table. Each line is a row. Commas separate the columns. It carries no layout, no motion, no fonts, no footage. It is data and nothing else. That is exactly why it is useful here, and exactly why “convert” is the wrong word.
Think about what you would need to do by hand to make one video from one row. You would open a design, type the name into a text layer, swap a photo, change a color, check that the text still fits, render, and name the file. Now repeat that for every row. A converter would have to guess your design, your fonts and your layout rules. It cannot. So the work splits cleanly in two. The template carries every design decision, made once. The data carries every difference, one row each.
In Dataclay’s system, each row in a spreadsheet or each object in a JSON array is one render job. Each column or key is one dynamic element in the template. The cells can hold text, URLs, image file paths or hex color values. That is the whole contract. If you can lay your content out in rows and columns, you can drive a template with it.
Here is a made-up but typical case. A regional retailer has a CSV of store promotions. There is one row per store. The columns are the store name, the city, the headline offer, a link to a store photo, and a brand color for that region. The marketing team wants one short promo video per store for social channels. Nobody wants to build that many videos by hand, and nobody wants to explain to a designer that the same layout needs to be repeated that many times.
With a rigged template, the designer builds the promo once. Layers for the store name, the city, the offer, the photo and the accent color are marked as dynamic. The CSV, in a form Templater can read, supplies the values. Templater merges row one, renders, merges row two, renders, and keeps going. The designer’s time goes into the one master design. Adding more rows takes little extra design effort.
This is the honest picture of where the effort lives. Rigging the template takes real thought. The data takes care. The rendering takes machine time that nobody has to watch. If a tool promises that no template is required, ask what the output will look like, because the layout choices have to come from somewhere.
One more thing a CSV cannot do is judge. It will happily hold an offer line that is far too long for the space, a photo link that is dead, or two rows with the same name. The template can absorb some of that. Templater’s layout rules adjust dynamic layers relative to their containing composition and sibling layers, so a longer headline can push things around instead of overflowing. Its Time Sculpting rules crop, trim, shift and stretch layers, which helps when footage clips differ in length. Neither one can fix a missing file. Those problems belong to the data, and the data is your job to check before you press go.
How a CSV gets into Templater or Que
This is the question underneath the search, and it has a slightly awkward answer. Templater reads its data from a short list of sources. Per Dataclay’s documentation, a Templater-compatible project can use a Que campaign, a native Google Sheets document, a tab-separated value file, a plain text file containing JSON, or remote JSON from a URL feed. A plain .csv file is not on that list. A tab-separated file is.
That sounds like a small gap, and mostly it is. A TSV is the same table with tabs where a CSV has commas. You can export one from Excel, Apple Numbers or Google Sheets. If your CSV came out of a database or a CRM, open it in a spreadsheet and save or export it as tab-delimited text. Our post on making After Effects videos from an Excel file walks through that export step by step. Templater also reads a Google Sheet directly, so another route is to paste your rows into a sheet and link it, which skips the export entirely.
Two details matter. First, TSV and JSON text files must be UTF-8 encoded so that Templater reads international characters correctly. If your store names include accents or non-Latin scripts, check the encoding before you blame the template. Second, Templater reads text as it appears in the spreadsheet, so commas, quotes, apostrophes and new lines inside a cell are accepted. Right-to-left scripts are supported too, as long as the installed fonts have the glyphs. That is one reason a TSV is friendlier than a CSV for messy marketing copy. A comma inside an offer line cannot be mistaken for a column break when the separator is a tab.
The second path is Que. Que is Dataclay’s web-based job manager, built for Templater users. It stores jobs in campaigns, and it has an import that accepts the file types people actually have. In the Que Dashboard, you open a campaign, choose Batch Import, and drop in a file. The supported types are a JSON file holding a single array of objects, a Microsoft Excel file with a header row, and CSV or TSV files. You pick a satellite to process the jobs and an autograf to supply the template, then click Upload. When the import finishes, each row shows up as a job in the campaign.
So if the file you have is a CSV and you would like to keep it a CSV, Que is the path that takes it as it is. Developers can do the same thing in code through Que’s Import Jobs endpoint. It takes a campaign ID and an array of job objects. If a job leaves out the aep or bot keys, Que fills in the campaign’s default autograf and satellite. If a job leaves out render-status, Que sets it to ready. Our post on inserting a batch of jobs into Que covers the dashboard steps in more detail, and the post on data sources in Templater compares the options side by side.
Which path fits? Use a Google Sheet if several people edit the content and you want changes to appear at the next preview. Use a TSV if the data is a one-time export and you want nothing in the cloud. Use Que if the volume is large, if you want jobs tracked in one place, or if another system will add rows later. Que is also where the pay-as-you-go option lives. Dataclay describes Que Meter as a way to produce customized videos without a standing subscription, aimed at occasional or experimental campaigns. See the Que product page for how it works and our licensing page for pricing.
A word on editions, because it changes what you can do. Templater comes in Rig, Pro and Bot. Rig prepares the project for dynamic content and lets you preview how layers react to data. Pro adds batch rendering and batch replication of re-versioned compositions from your data source. Bot adds the automation layer, which checks for incoming data and starts renders without a person, and it includes command line control. For a one-time batch from a CSV, Pro is the natural fit. If new rows will arrive over time and videos should appear on their own, look at Bot. The free trial contains all features, with a watermark added to the dynamic layers, so you can try the whole path on your own file before deciding.
Templater editions
Rig, Pro, and Bot do different jobs
| Edition | What it adds | Fits this scenario |
|---|---|---|
| Rig | Prepares project; previews layer reactions to data | Design and preview only |
| Pro | Batch rendering and batch replication from data source | One-time batch from a CSV |
| Bot | Automation layer; checks for incoming data; command line control | New rows arrive over time, videos appear on their own |
Build the file and the template so row 212 does not break the batch
Most batch failures are not render failures. They are data failures that show up late. You can catch nearly all of them with some discipline in the file and a few small habits in the template.
Name your columns to match your layers. Templater maps data to layers by name. A text layer called storeName in the template takes its value from a column called storeName. Dataclay’s guidance is to give every dynamic layer a unique, easily recognized name with no spaces, and to rename the layer explicitly even if its source name already looks right. If you skip that step, Templater may not treat the layer as dynamic. Apply the Templater Settings effect to each dynamic layer once, and only once, because applying it twice produces unexpected results.
Do not use reserved names for your own columns. Templater reserves a handful of column names. These are id, target, output, aep, render-status, bot, module and render-settings. They are case-sensitive, and using one as a dynamic layer name can cause an error when the data source is processed. Several of them are useful on purpose. The output column sets the file name for each render. The target column picks which composition to render, which lets one sheet feed several aspect ratios. The aep column switches between project files row by row. Use them deliberately instead of by accident.
Give every row a unique id or output name. This one deserves its own paragraph because the cost of getting it wrong is silent. When the Bot is rendering and two rows resolve to the same output file name, the later render overwrites the earlier one. Nothing warns you. You end up with fewer files than rows and no obvious reason. Dataclay’s own docs call this out, and the fix is a column of unique values. If you leave out the output column, Templater builds names from the id or the row index. Setting your own is better, because you will need to find a specific store’s video three weeks from now. Our post on designing a canary row shows one way to test the naming logic before a full run.
Check the file for the problems a machine cannot fix. This is general practice, not a Templater feature. Before any run, scan your columns for empty cells where the template expects a value. Look for links that do not open, text far longer than the design can hold, and duplicate rows. Sort each column and read the shortest and longest values, because the extremes are where layouts break. Fix the file first. It is far cheaper to correct a cell than to re-render a video.
Get footage and image references right. Cells can hold image file paths or URLs. Local paths are resolved relative to the footage folder you set in the Templater panel. Cloud links need to be reachable by the machine that renders. Dataclay documents several ways to use Google Drive files as remote footage, including a media connector and the Google Drive desktop app that mounts a drive on the rendering machine. A missing file is one of the more common reasons a single row fails, and Templater reports media problems under its ERR::MEDIA status instead of leaving you to guess.
Preview before you commit. Templater lets you cycle through composition previews with your data merged in, so you can see how each row will look before any render starts. Spot checks go further. They render a quick sample across many jobs so you can see how the data merges. Use them. Our post on preview modes and spot checks explains what each one shows. Scrub a few rows from the start, a few from the end, and the rows with the longest text. If those look right, the middle usually does too.
Know what the statuses mean. Each job moves through a render-status. Ready means the job has its data and has not been queued. Queued means it is waiting for a machine. Processing means Templater is actively rendering it. Done means the render finished. Canceled means it was stopped. Error codes such as ERR::MEDIA for asset problems and ERR::AFXHOST for After Effects errors mark failures. If you use Que, you can watch these in the campaign’s job list, and our post on Que job states explains what to do with each one. In a TSV or a sheet, Templater writes status back to the render-status column, which is why that column should only ever contain ready or be empty when you edit it.
Go back to the retailer. The team exports the store sheet to a TSV, names the columns to match the layer names, adds a unique id for each store, and previews the longest city name and the shortest offer. They find that one city name wraps to a second line and collides with the logo. They adjust the layer’s layout rule in the template. They rerun the preview, then start the batch. That fix is quick before the batch. The same fix after the batch would have meant re-rendering every affected video.
When this is the wrong tool, and where to go from here
This approach is not for everyone, and saying so up front saves you a trial.
It is the wrong tool if you do not have a design. Templater works inside Adobe After Effects. It assumes someone can build or buy an After Effects template. Dataclay’s Autografs are pre-built After Effects compositions already rigged with Templater and placeholder data, which can shorten that step, but you still need After Effects to run it. If your team does not use After Effects and has no wish to, a browser-based video tool may suit you better. Compare it honestly against your volume, your brand rules and how much control you need over motion.
It is the wrong tool if you need a different video structure for every row. Templater is strong at the same design with different content. If row one needs a thirty-second story and row two needs a completely different edit, you are describing editing, not data merging. The data can change text, images, colors, footage and language. It does not decide what a good edit is.
It is the wrong tool for a handful of videos. If you need five versions, make five versions. The rigging and data work pays off as the count grows. Our post on when automating is worth it is a fair place to check your own numbers, and the honest answer for small counts is often to just do it by hand.
It is also not a substitute for generative video. A template renders exactly what you designed. It will not invent a scene. If you want a model to generate imagery for each row, that is a different kind of tool with different risks. The two can work together, since Templater 3.7 and later can use a connected ChatGPT or Claude account to generate text for layers that have no content in the data source. Dataclay describes that as most appropriate for placeholder text while you design and preview, and says it may or may not suit production copy. Read your generated lines before they ship.
If the approach fits, the next steps are short. Get a rigged template, either your own or an Autograf. Prepare a small sample of your CSV, ten rows including your ugliest ones, as a TSV or in a Google Sheet. Run the preview, fix what breaks, and render that sample. Only then load the full file. If you plan to use Que, start with the same ten rows as a batch import and watch the jobs move from ready to done. Dataclay’s trial lets you do all of this with a watermark on the dynamic layers.
The bottom line is simple. A CSV does not become video by itself, and anything that claims otherwise is hiding the template. What you can build is a pipeline where design happens once, data changes freely, and the machine does the repetition. For a few hundred rows and an existing After Effects workflow, that pipeline is the practical answer, and the only real obstacle is making sure the file is clean before you begin.
Before you pick a path, a few questions come up almost every time. Here are the ones we hear most from people with a spreadsheet and a deadline.
Frequently Asked Questions
Can Templater read a CSV file directly?
Not as a data source on its own. Templater reads a Que campaign, a native Google Sheet, a tab-separated value file, a plain text JSON file, or a remote JSON feed. A CSV is one small step away. Open it in a spreadsheet and export it as tab-delimited text, or paste it into a Google Sheet. Que is the exception, because its batch import accepts CSV files as they are, alongside TSV, Excel files with a header row, and JSON arrays.
Do I need After Effects to turn a spreadsheet into videos this way?
Yes. Templater is an After Effects plugin, and the template it drives is an After Effects project. If you do not have a template, Autografs are pre-built After Effects compositions already rigged with Templater and placeholder data. You still need After Effects to run the render. If your team does not use After Effects, this is probably not the right workflow for you.
What file encoding should my data use?
Use UTF-8 for tab-separated and JSON text files. Dataclay’s documentation says this is required so Templater can read international characters correctly. If accents or non-Latin letters show up garbled in a preview, check the encoding of the file before you change anything in the template. Text inside cells can include commas, quotes, apostrophes and new lines, and right-to-left scripts are supported when the installed fonts have the glyphs.
Why did I get fewer video files than I have rows?
The most likely cause is duplicate output names. When two rows resolve to the same file name, the Bot overwrites the earlier render with the later one, and it does not warn you. Add a column of unique values and use it as your id or output name. Also check for rows that ended in an error status such as ERR::MEDIA, which means Templater could not find or import a media asset. Both are visible in the data and in the job list.
Should I use a Google Sheet, a TSV file or Que?
Use a Google Sheet when several people edit content and you want changes to appear at the next preview. Use a TSV for a one-time export with nothing in the cloud. Use Que when the volume is large, when you want jobs tracked in one place, when another system will add rows later, or when you want to import a CSV without converting it. All three feed the same kind of template, so you can change your mind later without redesigning anything.
Can I try this on my own file before buying?
Yes. The free trial contains all features for all versions, with a watermark added to the dynamic layers. Start with ten rows of your real data, including the longest text and the odd cases, and run a preview and a small render. That tells you far more than a demo with clean sample data, and you can see the licensing options on our product pages afterward.
