Data to AI™ / Gemini/Microsoft Ads
Pull Microsoft Advertising keywords, ads and hourly results into the terminal or IDE, and let Gemini write the audit files, notebooks and pre-deploy checks your team keeps in the repo.
Better answers, fewer tokens.
Gemini CLI
Which campaigns lost the most impressions to budget last month, and were they profitable?
Brand Search lost 38% of eligible impressions to budget while returning 6.2 per unit of spend, so it is the clearest case for more budget. Generic Search lost less and returned far less.
| Campaign Name | Spend | Impression Lost To Budget Percent | Return On Ad Spend |
|---|---|---|---|
| Brand Search Campaign Performance Report | $4,100 | 38% | 6.2 |
| Generic Search Campaign Performance Report | $9,800 | 21% | 1.9 |
| Performance Max Campaign Performance Report | $6,300 | 12% | 3.4 |
All three columns come from one Campaign Performance Report with Monthly time aggregation, so the shares are not averaged from daily rows.
Microsoft Advertising (formerly Bing Ads) runs paid search campaigns across Bing and Microsoft's partner network. Dataddo extracts its reports, from account and campaign down to keyword, search term and ad, with the attributes and metrics you pick for each report type.
Query the Keyword Performance Report for the last 30 days and write every Keyword with Quality Score below 5, its Expected Ctr, Landing Page Experience and Spend to audits/msads-quality.csv.
A technical marketer runs the same prompt every month from the terminal. The CSV is committed next to the previous ones, so the diff shows which keywords improved after landing page work.
From the Microsoft Ads source with Hourly time aggregation, load Clicks, Conversions and Spend for the last four weeks into notebooks/dayparting.ipynb and plot them by hour.
An analyst in VS Code uses Gemini Code Assist in agent mode to build the notebook, then proposes an ad schedule based on the hours where conversions cost the least.
List each Final Url from the Ad Performance Report with Spend in the last 7 days and compare them with the routes in src/pages. Flag any URL we are about to remove.
A developer runs the check before merging a site restructure, so paid traffic is not sent to pages that will return 404 after the deploy.
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 | |
|---|---|---|---|
| Audits under version control | Quality Score exports are downloaded to laptops and overwritten, so nobody can see how a keyword changed over months. | Gemini CLI writes each audit from the Keyword Performance Report into the repository, where history and review come with git. | Audit trail in the repo |
| Granular time data | Hourly data has to be exported separately and reshaped before a notebook can plot it. | A source with Hourly time aggregation is queried by field name from the IDE, so the notebook starts from clean columns. | Dayparting without reshaping |
| Ads and code in one place | Developers do not know which landing pages receive paid clicks until traffic drops after a release. | Gemini checks Final Url values with recent Spend against the codebase before the deploy, in the same terminal session. | No paid clicks to dead pages |
| Token consumption | Gemini spends tokens exploring raw Microsoft 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 Microsoft Ads 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
|
|---|---|---|
| Campaign Performance Report |
|
Which Campaign Name saw the biggest week-over-week drop in Conversions? |
| Keyword Performance Report |
|
Which keywords collect many Clicks but have the highest Cost Per Conversion? |
| Search Query Performance Report |
|
Which new search terms appeared this week with more than 20 Clicks? |
| Ad Performance Report |
|
Which Headline wording do our best-converting ads share? |
| Budget Summary Report |
|
Which Campaign Name had the biggest jump in Daily Spend this week? |
| Professional Demographics Audience Report |
|
How did Conversions by Industry Name change between the last two months? |
| Geographic Performance Report |
|
Which countries improved their Return On Ad Spend most quarter over quarter? |
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 a Microsoft Ads source in Dataddo and attach it to a Google Gemini destination or AI Model. In ~/.gemini/settings.json, add the dataddo server under mcpServers with httpUrl https://headless.dataddo.com/mcp-data, restart Gemini CLI and run /mcp auth dataddo to sign in. Gemini Code Assist in VS Code reads the same file.
The Dataddo setup covers Gemini CLI and Gemini Code Assist. Gemini CLI reads ~/.gemini/settings.json or a project's .gemini/settings.json, VS Code agent mode reads the same file, and IntelliJ uses the mcp.json of Gemini Code Assist.
With the url key Gemini CLI falls back to SSE, which Dataddo does not support. Use httpUrl with no headers or oauth block, restart Gemini CLI and run /mcp auth dataddo.
Yes. Once signed in, Gemini reaches every flow or Context attached to an AI destination or AI Model in that Dataddo account, not only the Google Gemini ones. Keep flows Gemini should not query off AI destinations and AI Models in that account.
Gemini CLI asks you to confirm each Dataddo tool call, and you can choose Always allow per tool. Queries are read-only through the semantic layer; Dataddo never accepts SQL from the model.
Credentials are stored in ~/.gemini/mcp-oauth-tokens.json and renewed automatically. If the session expired or was revoked, run /mcp auth dataddo again and sign in with your Dataddo email and password.
Gemini CLI opens the Dataddo sign-in page in a web browser, so it needs a browser on the same machine. Google and Microsoft SSO are not offered for this sign-in; use your Dataddo email and password, with MFA if prompted.
