Data to AI™ / Claude/X Ads
Ask Claude which promoted posts, campaigns and messages earn their budget on X, using ad data Dataddo extracts on schedule before the 7-day API window closes.
Better answers, fewer tokens.
Claude
Which X Ads campaigns got the cheapest link clicks last week?
Spring Webinar Promo had the cheapest link clicks at $0.41 each, on less than a quarter of the spend. Brand Awareness Q2 took the most budget but cost $1.12 per URL click.
| Campaign Name | Spend | URL Clicks | Impressions |
|---|---|---|---|
| Spring Webinar Promo Campaign Analytics | $1,230 | 3,000 | 184,500 |
| Product Launch Retarget Campaign Analytics | $2,050 | 3,420 | 142,800 |
| Brand Awareness Q2 Campaign Analytics | $4,480 | 4,000 | 611,200 |
Spend is Billed Charge Local Micro divided by 1,000,000, in the ad account's currency, for placements on X itself.
X Ads (formerly Twitter Ads) extracts paid and organic performance from an X ads account: campaign spend and engagement, promoted post results with web conversions and video completion, post text, and campaign budgets and status.
Match Promoted Tweets Metadata to Promoted Tweets by Tweet ID. Which Full text themes bring the highest Conversion Purchases Sale Amount per unit of Spend?
Claude reads the post wording next to its results, so the social lead writes the next round of promoted copy from patterns that paid off instead of from the post everyone liked internally.
Every Monday, query Campaign Analytics for the last 7 days through the dataddo server and append Campaign Name, Date Start, Spend and URL Clicks to data/x-ads/campaigns.csv.
In Claude Code, an analyst keeps a growing history in version control. Because X only allows extraction up to 7 days back, the file becomes the long-term record for year-on-year questions.
Compare Engagements per Impressions for Organic Tweets and Promoted Tweets over the last 7 days. Which organic posts did better than our promoted ones?
Claude shows organic posts that outperform paid ones, which become candidates for promotion, and promoted posts that would not earn attention without budget.
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 | |
|---|---|---|---|
| Copy next to results | Post text sits in one export and promoted results in another, so someone matches rows by Tweet ID by hand before Claude can compare them. | Claude queries Promoted Tweets Metadata and Promoted Tweets through the semantic layer and joins them on the ID for you. | Copy and results in one answer |
| History beyond seven days | The X API only returns data up to 7 days back, so a question about last quarter has no data unless someone exported it at the time. | Scheduled extraction keeps every run, and Claude Code can write a dated archive into the repository for longer comparisons. | No lost weeks |
| Money in micros | Raw exports show Billed Charge Local Micro and Daily Budget Amount Local Micro, which read a million times too high. | The Spend field in the analytics datasets is already divided by 1,000,000, and Claude reads field descriptions from the semantic layer. | Spend in real currency |
| Token consumption | Claude spends tokens exploring raw X Ads 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 X Ads datasets marketing teams use most, with their real field names. Claude queries them by name through the Dataddo semantic layer.
| Dataset | Metrics and attributes |
Ask Claude
|
|---|---|---|
| Campaign Analytics |
|
Which Campaign Name had the lowest Spend per URL Clicks over the last 7 days? |
| Promoted Tweets |
|
Which promoted posts returned the highest Conversion Purchases Sale Amount for their Spend? |
| Promoted Tweets Metadata |
|
Which promoted posts mention a discount in their Full text? |
| Campaigns |
|
Which campaigns have an Entity Status of paused but an End Time still in the future? |
| Organic Tweets |
|
Which organic posts had the most Engagements per Impressions in the last 7 days? |
| Account Analytics |
|
How did account-level Impressions and Engagements develop day by day this week? |
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 →Create an X Ads source in Dataddo with the X v1 authorizer and datasets such as Promoted Tweets and Campaign Analytics, then attach it to a Claude destination or AI Model. In Claude, open Customize > Connectors, add Dataddo Data Access from the directory and sign in. Claude Code picks it up too.
Yes. Dataddo lists it as X Ads (formerly Twitter). Datasets keep the Twitter-era names, such as Promoted Tweets and Organic Tweets, so Claude finds them under those names.
Only as far as you have extracted. The X API allows extraction up to 7 days back, so historical data cannot be loaded later. Start the source early and let it run on schedule to build history.
X v1 is recommended; X Custom uses your own app credentials. The X Ads API needs its own authorizer, separate from X Organic, and a wrong one causes an Invalid callback url error.
No. Claude queries extracted data through the semantic layer by entity and field name. Dataddo never accepts SQL from the model, and nothing is written back to X Ads.
Yes. Signed in with your Claude account, Claude Code loads the Dataddo connector automatically, which you can check with /mcp. With an API key, add the server from your terminal and sign in.
On Pro and Max you add Dataddo Data Access yourself. On Team and Enterprise an Owner adds it once and every member signs in with their own Dataddo account.
