TrueMetrics is a data consulting firm based in Beijing that connects marketing strategy to data technology. Clients come to them with a business problem: work out which touchpoints deserve credit, sharpen creative strategy, predict churn. Nobody hires a consultancy to export data. But for years, exporting data is where the work started.
The challenge: two teams, two languages, half a project spent on plumbing
TrueMetrics begins every engagement with an audit, and the audits kept surfacing the same finding. Most clients had no standardized access to their own marketing data. Without that foundation, the interesting work stalled while the team moved, cleaned, and reconciled data.
Scarlett Zhang, Head of Consulting & Analytics at TrueMetrics, has spent more than a decade in data consulting. She traces the cause to a communication gap inside the client's own organization: "The marketing team speaks about ROI, attribution, and creative strategy. The IT team speaks about API documents, schemas, and databases.” Neither side could carry the work alone. Marketers couldn't read the API documentation. Engineers didn't know the marketing context well enough to decide what mattered. So the foundation work fell to TrueMetrics, and it consumed roughly half of a typical engagement. That's time clients pay for without wanting to, because it isn't what they came for.
The cost showed up in three places. Engineering time went into building API connections and maintaining databases. Learning time went into decoding how each advertising platform defines its own metrics and dimensions. Validation time went into checking exported numbers against the platforms, where differences in time zones, refresh frequency, and backfill behavior all had to be reconciled by hand.
The solution: a three-step delivery model
TrueMetrics rebuilt its delivery around Supermetrics' Marketing Intelligence Platform and a three-step sequence that now runs in two to three weeks.
- Integration. Ready-to-use connectors bring client data from Google Ads, Meta, TikTok, and Google Analytics into BigQuery on a schedule. Time zone handling, refresh frequency, and backfill are handled by the platform, so validation stops being a manual task. This step now takes two to three weeks, depending on how many sources a client needs.
- Reporting layer. With clean data already in the warehouse, TrueMetrics builds dashboards and BI architecture on top of it rather than building the data foundation and the dashboards at the same time. This covers market research and competitive analysis, omnichannel measurement, and creative and audience intelligence, all from the same data foundation.
- AI. Because the data arrives in a consistent structure, connecting a client's warehouse to an agent is a short exercise.
As a Google Cloud partner, TrueMetrics connects client warehouses directly to Gemini Enterprise and BigQuery AI, so client marketing teams ask questions of their own data in natural language.
In practice: a fast-growing D2C brand
One direct-to-consumer ecommerce client had grown from three advertising channels to more than a dozen media and social platforms, but had no unified view across them.
The marketing team was evaluating performance one platform at a time, with conversion data inflated — different platforms each claiming credit for the same sale — and no visibility into user drop-off across campaign types throughout the conversion funnel. Each platform was managed separately, so any cross-channel comparison meant pulling numbers by hand. Reviewing performance across channels had become a monthly or quarterly exercise.
Supermetrics automated the extraction and ingestion of multi-channel ad data and API connections directly into BigQuery, eliminating manual engineering overhead and ensuring real-time data accuracy. TrueMetrics then built a custom Data-Driven Attribution (DDA) model on top, combining Google Ads performance with on-site GA4 user behavior data, giving the client a full view from impression through to conversion for every channel. They then built a custom attribution model so the client could compare return across touchpoints.
The clean data foundation changed how the client thought about budget allocation entirely:
- 35%+ reduction in inflated order reporting
- 25%+ of conversions reattributed back to Google Ads — sales that had previously gone uncredited
- Google's contribution to 10% of total conversions had been previously underestimated
- 6X growth in upper-funnel branding campaigns, once the team had the evidence to justify the investment
Two things changed. Decisions got faster because performance data no longer waited for a reporting cycle. And the client gained enough confidence in the numbers to invest in upper-funnel campaigns, which are the hardest to justify when attribution credits whatever sits closest to the sale.
In practice: a gaming client that couldn't afford to rebuild
The second client was in the opposite situation. A gaming company with a decade in the market already had a mature data warehouse, and it was rigid. The marketing team wanted churn prediction and lifetime value models, but the features those models needed weren't in the warehouse, and restructuring it meant months of evaluation. In a competitive category, months is a long time to stand still.
TrueMetrics built a pilot around an architecture that required no data movement out of the client's environment. Supermetrics brought advertising platform data and key user behavior data into BigQuery. The models were built inside BigQuery, with no extra servers to provision. Gemini and BigQuery AI were connected directly, so the marketing team could work on feature engineering in natural language and get insights back without a data scientist in the loop.
The pilot ran for two to three weeks. It added capability to the existing warehouse rather than replacing it, which mattered internally, since teams tend to be protective of the infrastructure they built. For this gaming client, that meant a churn and lifetime value pilot delivered in two weeks instead of the planned four, with 85%+ model accuracy and a 50% efficiency gain. Once the use case was validated, the client had a structure it could copy across other games and markets.
The client is now running a continuous optimization loop — an MLOps/LLMOps automated flow where data feedback drives model updates, policy adjustments, and effectiveness monitoring, without manual intervention at each step.
The result: foundations in weeks, not months
The headline change for TrueMetrics is where its time goes. Data integration, which used to consume roughly half the project timeline, now takes two to three weeks, and the capability transfers across the team rather than living with one specialist.
That shift also changed how the team thinks about scoping. When the deliverable was a dashboard, the project defined its dimensions and metrics up front. AI removed that certainty, because nobody knows in advance which question a client will ask.
For clients, that's the difference between a data project that ends at a dashboard and one that keeps producing answers.
Ready to build the data foundation your AI runs on?
See how Supermetrics brings your ad, social, and search data into BigQuery in a format your team and your AI agents can both work from, on a daily schedule you don't have to manage.