One report adds Google's Cost to Meta's Amount spent. The monthly deck uses a different Meta field entirely. The forecast was built by someone who left in March. All three are called "total spend", none of them match, and you find out in front of the client.It isn't a reporting problem.
It's a definitions problem — and it multiplies by every platform, every client, and every person who has ever built a card.
Spend, impressions, reach, clicks, CTR, CPM, CPC, conversions, conversion value and ROAS are mapped for you — along with the dimensions you group by: date, campaign, ad group, ad, placement, country, device and source.
The part that earns trust is what comes with the name. Each unified field carries the right behaviour, not just the right label. That's the difference between a number that looks right and a number that is right.
Blended CPA across three ad platforms. Spend after your management fee. Leads split into qualified and everything else. No platform will ever ship those — they're yours.
So describe the one you want the way you'd explain it to a colleague, and Jepto works out which fields that means. You see the number before you save, and Jepto checks the maths on the way in — so a broken definition never reaches a client report.
Cards reference fields by name — they never keep their own copy of the definition. Correct a formula, sharpen a description, and everything using that field reflects the change on the next load.
Descriptions do quiet work too. Whatever you write becomes the tooltip the next person sees when they hover the field — so "which spend metric do I use?" stops being a Slack thread. And scope is yours: build a field for one client while you're proving it out, then promote it account-wide once the team agrees it's the definition.
Three moments where a shared catalog either saves you or quietly embarrasses you. You've had all three.
Ask any AI what you spent on paid last month and you'll get an answer. Confident, formatted, rounded to the dollar, ready to paste into a client email. Whether it's right depends entirely on whether it knows what your metrics mean.
A broken script throws an error. A broken number just renders — and the only person who can catch it is someone who already knew the answer. Here's the same question, asked twice.
Writing a connector is a weekend job now. Keeping it correct is a subscription — and the worst API changes are the quiet ones. Nothing errors; a metric just starts meaning something slightly different, and your year-on-year comparison has been lying since January. Someone has to notice, and then migrate it. On Jepto, that someone isn't you — while the custom fields you wrote stay exactly where they are.
Most agencies already have a data model. It's a spreadsheet, a Notion page, or one senior person's memory — and it stops working the week they go on leave.
Every analytics tool with a chat box will tell you it's grounded in your data. These five answers tell you whether it actually is.
Everything you need to know about the Data Model.
Start asking your data smarter questions today. Connect your first data source in minutes.
