For data teams
Numbers you can trust, and check.
Two people ask for ROAS and get two numbers. Mediatape works out every figure before the AI sees it, with your definitions, so the same question gets the same answer. Each answer says what it is made of.
What’s our blended ROAS by store this month?
Using your saved metric Blended ROAS (purchase value ÷ spend), with a purchase counted the way your team mapped it in each account.
| Account | Counts as purchase | Spend | Blended ROAS |
|---|---|---|---|
| US store · Meta | Purchase | $12,400 | 2.9× |
| US store · Google Ads | Online sale | $6,100 | 3.4× |
| EU store · Meta | Purchase | $4,300 | 2.2× |
Oct 1–15. EU store converted from EUR at each day’s ECB rate. Google Ads has no purchase value for Oct 3–4, so its ROAS covers 13 of 15 days.
Why Mediatape
Same assistant. Different answer.
The assistant does the arithmetic either way. Mediatape decides what goes into it.
What’s our blended ROAS across all accounts this month?
Export spend and purchase value from each platform, convert them to one currency, then divide total value by total spend.
What’s our blended ROAS across all accounts this month?
2.9× across your three accounts, using your saved Blended ROAS and your purchase mapping. The EU store is converted from EUR at each day’s ECB rate.
Numbers you can trust
Every answer says what it is made of.
- Dates and freshness
- Which days are included and when the data was last collected.
- Reporting definitions
- Which conversion and attribution window a number uses. Original figures are kept as reported.
- Currency
- Every amount is labelled with its currency. Accounts in different currencies are converted at each day’s ECB rate.
- Missing data
- Gaps are shown. A broken connection is named before the answer.
- Creative labels
- Left blank when unsure. Your team’s corrections win.
- Private to your org
- Your creatives are analysed only for you. They are never compared across customers or used to train shared models.
How numbers are worked out
The AI asks. Mediatape does the maths.
Your definitions, applied the same way in every chat.
No SQL from the model
The AI asks a structured question: metrics, dimensions, filters and limits. Mediatape runs it, so the same question is worked out the same way every time.
Rates from totals
Cost per purchase, CTR and ROAS are recomputed from summed spend and results, never averaged across rows.
Read-only, one org at a time
Mediatape can’t change a budget or an ad, and every question reads only your org’s own data.
Example questions
Questions data teams ask.
- Which attribution window does this cost per purchase use?
- Which days are missing for our EU account this month?
- When was the Google account last synced?
- How is blended ROAS defined for our org?
- Which creatives have no label, and why?
- Spend by account in USD, with the exchange rates used.