Data to AI™ / Gemini/Facebook Organic
Turn transposed Page insights, reviews and lead form counts into tidy files and scripts from your terminal or VS Code, with personal columns left out when you create the source.
Better answers, fewer tokens.
Gemini CLI
How much of our Facebook Page media views were organic in September?
Organic media views held steady at 177k to 195k a week. Paid views jumped in the week of 8 Sep, when posts were boosted, so that spike is not organic growth.
| Week | Page Media View (Organic) | Page Media View (Paid) |
|---|---|---|
| 1-7 Sep Page Daily Media Views by Ad Status | 182,400 | 96,300 |
| 8-14 Sep Page Daily Media Views by Ad Status | 176,900 | 141,200 |
| 15-21 Sep Page Daily Media Views by Ad Status | 188,100 | 88,700 |
| 22-28 Sep Page Daily Media Views by Ad Status | 194,600 | 90,500 |
Organic and paid views are separate rows from Facebook's is_from_ads breakdown, added up by week.
Facebook Organic collects non-paid data from your own Facebook Pages: post and reel insights, page metrics, fans by market, comments, reviews, inbox metrics and lead form submissions.
Query Page Daily Media Views by Ad Status for Q3 and write scripts/pivot_media_views.py that turns the Title rows into organic and paid columns per Date.
A data engineer replaces the monthly spreadsheet pivot with a short script in the analytics repo. Gemini CLI checks the real Title values before writing it, so the column mapping matches what Facebook returns.
Pull Created time, Recommendation type and Review text from Page Reviews for this year and save them to data/page-reviews.csv for our sentiment notebook.
An analyst runs the team's own text model over the reviews in a notebook. Each refresh is one prompt in Gemini Code Assist, and the file always holds the latest scheduled extraction.
Count rows in Leads Transposed per Form Id and Campaign Name for last week and append them to reports/leads-weekly.csv. Leave Email out.
RevOps keeps a running file of lead volume per form. Email is unticked in the source, so the terminal session never handles it, even if a prompt asks for the column.
Data to AI™ is priced per seat, like the AI assistants your team already uses, with ~25% off on annual billing. Need tailored contracts or volume pricing? Talk to us about Enterprise.
Serve governed business data to Claude, ChatGPT and other AI tools over MCP.
$10.49/month
per seat
3 seats minimum
Flexibly deploy Dataddo in any cloud or hybrid environment.
Custom
We have a payment model that works for you
| Without Dataddo |
With Dataddo
|
Outcome for you | |
|---|---|---|---|
| One row per metric | Transposed exports are pivoted by hand in a spreadsheet every month, and a renamed metric silently breaks the formulas. | Gemini Code Assist writes the pivot once, reading Title and Metric Value through the dataddo server, and the script runs again on new data. | A pivot you can rerun |
| Personal data from lead forms | A full lead export with names, emails and phone numbers lands in a project folder on a laptop. | Untick columns such as Email when you create the source, so they never reach the flow and Gemini cannot read them. | No plain-text emails locally |
| Credentials on the machine | Long-lived Page access tokens get copied into .env files across several machines and repositories. | Gemini CLI keeps the Dataddo sign-in in ~/.gemini/mcp-oauth-tokens.json and renews it automatically; /mcp auth dataddo signs you in again when needed. | No Facebook tokens in .env |
| Token consumption | Gemini spends tokens exploring raw Facebook Organic columns, sampling rows and retrying until it understands the data. | Field definitions and relations are supplied up front, so the model skips the exploration and goes straight to the answer. | Fewer tokens per answer |
| Answer quality | Prone to wrong joins, invented fields and numbers that are hard to check. | Answers grounded in defined fields. In the Dataddo benchmark, 78.5% of questions were answered correctly, against 65.5% with plain CSV files. | Better answers |
The Facebook Organic datasets marketing teams use most, with their real field names. Gemini queries them by name through the Dataddo semantic layer.
| Dataset | Metrics and attributes |
Ask Gemini
|
|---|---|---|
| Posts Lifetime Insights by Post |
|
Export Message, Lifetime Reach and Clicks for last quarter's posts to a CSV so a notebook can tag topics. |
| Page Daily Media Views by Ad Status |
|
Pivot Metric Value by Title and Date for last month into a CSV with one column per Title. |
| Page Daily Fans by Country |
|
Write weekly Fans Count per Country for this year to a CSV in the data folder of the repo. |
| Post Comments |
|
Which comments with a Comment Count above zero started threads last week, and what did their Message say? |
| Page Reviews |
|
Count reviews per Recommendation type for each month by Created time and return a markdown table. |
| Page Messages Lifetime Insights |
|
Return Displayed message response time and Unread message count per Page name as JSON for our support dashboard. |
| Leads Transposed |
|
How many leads per Form Id arrived each day by Created Time last week, without listing any Email? |
We asked 43 pre-registered questions of three real datasets - Google Search Console, HubSpot deals and Google Ads - and delivered the same rows to the same model in four ways. Only the data layer changed.
of questions answered correctly - the best of the four delivery methods tested
fewer confidently wrong answers - 14.0% of runs against 24.0% and 26.3%
more correct answers on messy CRM data - better on 8 questions, worse on none
Same model, prompts and rows in every condition; 759 graded runs, ground truth frozen before any run. On Google Search Console alone, documented CSV files did as well as Data to AI™ - the advantage comes from messier data such as HubSpot deals.
Read the benchmark →Add a Facebook Organic source in Dataddo with datasets such as Page Daily Media Views by Ad Status and Page Reviews, then attach it to a Google Gemini destination or AI Model. In settings.json, register the dataddo server with httpUrl https://headless.dataddo.com/mcp-data, restart Gemini CLI and run /mcp auth dataddo.
Yes. Transposed datasets such as Page Daily Insights Transposed hold each metric as a row with Title and Metric Value, and Gemini can pivot them in a script. Posts Lifetime Insights by Post already has one row per post.
Not if you leave them out. Untick columns such as Email or Phone Number in the Leads datasets when you create the source; they never reach the flow, and counts per Form Id still work.
Yes, with Page Daily Media Views by Ad Status. It uses Facebook's is_from_ads breakdown to store organic and paid views as separate rows, which a script can compare week by week.
Every flow or Context attached to an AI destination or AI Model in the Dataddo account you signed in with, including ones set up for other AI clients. For several Pages in one source, ask the Dataddo Solutions team about multi-account extraction.
Make sure the entry in settings.json uses httpUrl and not url, which falls back to SSE that Dataddo does not support, and has no headers or oauth block. Restart Gemini CLI and run /mcp auth dataddo.
Up to 2 years, the Facebook API limit, for datasets with a date range. Lifetime and metadata datasets always pull what is currently available. Load older months once with a full data re-sync.
