Data to AI™ / Claude/Google Ads
Ask Claude where search budget is lost, which keywords cost too much per conversion and which ads need rewriting, using the Google Ads reports you already sync.
Less tokens, better answers.
Claude
Which search campaigns lost the most impression share to budget last month?
Competitor terms lost 44% of its eligible search impressions to budget and Generic - Running shoes lost 31%. Brand - Exact lost only 2%.
| Campaign - Name | Cost Micros (Spend) | Search Impression Share | Search Budget Lost Impression Share |
|---|---|---|---|
| Competitor terms Campaign Report | 5,410 | 38% | 44% |
| Generic - Running shoes Campaign Report | 12,880 | 54% | 31% |
| Brand - Exact Campaign Report | 3,240 | 91% | 2% |
Cost Micros (Spend) was divided by 1,000,000 to match the amounts in the Google Ads interface.
Google Ads data comes as Reports with Metrics, Reports without Metrics and Keyword Planner Historical Metrics. You pick a report type such as campaign, keyword, search term or shopping, then the attributes, segments and metrics to extract.
Which campaigns had the highest Search Budget Lost Impression Share last month? Show Cost Micros (Spend) in currency next to Conversions.
A search lead sees where impressions are lost to budget rather than rank, and which of those campaigns convert well enough to deserve more money.
Query the Campaign Report through the dataddo server, convert every micros field to currency and write a weekly summary to reports/google-ads.md.
A marketing engineer uses Claude Code so the weekly report divides Cost Micros (Spend) and budget amounts by 1,000,000 the same way every time, reviewed in a pull request.
Which responsive search ads have the weakest Ad Group Ad - Ad Strength, and what do their Ad Group Ad - Ad - Responsive Search Ad - Headlines say?
Claude lists the weak ads with their headlines and drafts alternatives in the same chat, so the copywriter starts testing from a concrete shortlist.
Data to AI™ is free while in private pre-release. We onboard teams in waves and shape the product with their feedback. At launch, it will be priced like the AI assistants your team already uses. Need tailored contracts or volume pricing? Talk to us about Enterprise.
Serve governed business data to Claude, ChatGPT and other AI tools over MCP.
Free during pre-release
Flexibly deploy Dataddo in any cloud or hybrid environment.
We have a payment model that works for you
| Without Dataddo |
With Dataddo
|
Outcome for you | |
|---|---|---|---|
| Costs in micros | Pasting API output into Claude shows costs a million times higher than the Google Ads interface, and nobody notices until a summary goes out. | Cost Micros (Spend) carries its description from the semantic layer, so Claude knows to divide it by 1,000,000. | Costs match the interface |
| Performance Max coverage | An ad group export silently leaves out Performance Max, because Google reports those campaigns only at campaign level. | Attach a Campaign Report source and Claude covers Performance Max alongside the keyword and ad level reports. | No missing campaigns |
| Code and chat on one source | Analysts ask questions in Claude while engineers maintain separate API scripts, and the two sets of numbers drift apart. | Claude and Claude Code read the same Google Ads entities through one connector URL; Claude Code keeps its credentials in the system keychain. | Same numbers everywhere |
| Token consumption | Claude spends tokens exploring raw Google 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. | Less 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 Google Ads datasets marketing teams use most, with their real field names. Claude queries them by name through the Dataddo semantic layer.
Reports with Metrics read a relative date range, for example yesterday, that moves forward on every scheduled run; older periods are loaded with a full data re-sync. Reports without Metrics pull all current data on each run.
Google Ads connector →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 Google Ads sources in Dataddo for the report types you need, such as Campaign Report, Keyword View and Ad Group Ad, and attach them to a Claude destination or AI Model. In Claude, add the custom connector https://headless.dataddo.com/mcp-data and sign in. In Claude Code, run claude mcp add, then /mcp.
The Google Ads API reports costs in micros. Fields such as Cost Micros (Spend) must be divided by 1,000,000, so ask Claude to convert them in your prompt.
Google only reports Performance Max at campaign level. If your source extracts ad group or ad level data, add a Campaign Report source so Claude can see them.
Yes. Run claude mcp add with the Dataddo URL and sign in with /mcp. The credentials stay in your system keychain and Claude Code queries the same Google Ads entities.
No. Claude queries data Dataddo has already extracted, by entity and field name, and never writes to your Google Ads account.
Yes. Use multi-account extraction to pull the same report from several accounts, attach the flow to Claude, and ask Claude to group results by account.
As far as the flow holds. Reports with Metrics read a relative date range, so load older periods into the flow with a full data re-sync.