Google Ads to JSON: Fields to Check Before You Reconcile

Brand promotion: this article introduces Guangsuan's Google Ads management services.

If you already move social data to CSV or JSON for audit work, the Google Ads request feels similar: pull spend and results, shape it as JSON, then match it to the CRM. The mismatch that follows is rarely about JSON syntax. It comes from what was left out of the file — an unstable key, a mixed date rule, or a different idea of what counts as a conversion.

Direct answer: The Ads interface does not offer a JSON download option. To get JSON, call the REST form of the Google Ads API, or stage the data in a warehouse and produce JSON there. Before matching to CRM, write down the account time zone and range, keep stable IDs on each row, fix whether you count by interaction time or by conversion time, mark primary actions, and normalise micros once.

What the download menu actually gives you

The download control above a statistics table offers spreadsheet and document choices such as Excel CSV, TSV, PDF, XLSX, XML and Google Sheets. JSON is absent from that list. XML sits closest to the request, yet it serves XML readers rather than a JSON pipeline.

Turning a CSV or XML download into JSON on your own machine is allowed, but treat that script as part of reporting: keep it in version control, note who owns it, and test it against a small slice you have checked manually.

Choose the JSON route that fits your workload

API REST for code and pipelines. The platform exposes both gRPC and REST, and the REST form returns the same objects as JSON. You ask in GAQL and receive rows that carry only the attributes, metrics and segments named in that query. Access depends on a developer token plus an authenticated client, with tiered daily limits starting from test-account use. Credentials belong in a secret store, never in a shared report. For field names and limits, see Guangsuan's guide to exporting Google Ads data to JSON.

Warehouse staging for repeated joins. A BigQuery transfer or reporting connector can hold the data first and emit JSON later. That adds something to maintain, so it suits a recurring CRM match more than a single file. A one-off hand-off is usually faster as a local CSV conversion.

A short lock-down list before the CRM join

Start by noting the account time zone, currency and target dates on the file itself. An export stamped in the account zone can place a late-month lead in another month beside CRM timestamps.

Then keep keys that survive edits. Retain the customer identifier, retain the numeric campaign identifier beside its display label, and retain the breakdowns that keep rows distinct, including day, device, network slice and conversion action where used. Labels get renamed; numbers do not.

Reports built on session counts, interaction timing or conversion timing should stay separate. Click-time and conversion-time placements can put one purchase in different months, and both are legitimate until they are blended into one comparison. Pick one rule, note the window, counting choice and attribution setting, and apply it on both sides.

A reader with view-only rights can fetch reports already shared with them, so permission alone does not make a file joinable. Grain and counting rules still need the written note above.

Conversions and money: two places gaps appear

Only actions flagged as primary flow into the Conversions column used for bidding. Page views, contact taps and similar micro steps stay secondary and should stay labelled that way in the file. When sales numbers come from offline qualification, that file reflects only what the ad system could record.

Cost arrives in micros under the account currency, so apply the divide-by-one-million step a single time and record where it happened. Files in different currencies should not have raw totals summed; keep them apart or state a dated rate.

Reconcile narrow, then decide about bids

  1. Select one primary action, one campaign set and one closed period.
  2. Align the time rule, range and zone on both extracts.
  3. Isolate ad-sourced records from organic and referral entries.
  4. Inspect counting behaviour and duplicate loads before treating a remainder as unexplained.

As a hypothetical teaching illustration with invented figures, not a client outcome: a file logs 58 ad-recorded contacts in a month. Sales accepts 19 as qualified under a written definition, 14 dated within that month and 5 dated after it, with 4 moving to order. Each total answers a separate stage, so no total refutes another. When few contacts ever qualify, inspect query intent and page fit before moving bids.

Guangsuan's team works from search terms, ads, landing pages and conversion tracking toward purchase demand rather than clicks alone. Scope and fees are set in a contract, with no fixed inquiry promise. Details sit on the Guangsuan Google Ads management service page. For a first review, bring your action list with primary flags plus one annotated extract showing basis and zone.

FAQ

Can a vanished keyword mean zero rows were hidden? Yes. Very low-activity rows can be omitted from an edited view, so confirm filters and granularity before assuming the term never matched.

Interaction time or conversion time at month end? Use the rule that fits the question, keep it identical on both sides, and file the window plus zone with the extract.

Does JSON that parses cleanly mean it is ready to attribute? No. Parsing says nothing about row completeness, granularity, units, date rule or which actions were summed.

Back to all insights