This section applies only if you currently use Traackr's Data Lake offering, which will eventually be sunset. This page is here to help you get started with your migration.
The Data Lake delivers files. The API answers questions. Almost every practical difference follows from that. If the Data Lake is a library delivered to your door every week, the API is a librarian you can ask for one specific book.
What changes
| Data Lake | API |
What arrives | Files on a schedule, whether you need them or not | Exactly what you ask for, when you ask |
Work involved | Data engineering: ingest, store, clean, aggregate | Software engineering: call, read, display |
Time to first chart | Slow: you ingest everything before showing anything | Fast: ask and display |
Aggregation | You compute most figures yourself | Campaign and creator totals arrive computed |
Scope | Everything, every time | One campaign, one creator, one platform: your choice |
History | Each export is kept for a limited time | None: see below |
Freshness | As of the export run, days or weeks old | The Traackr app's figures, within a few hours |
3 things that get much easier with the API
1. Many numbers arrive computed. The Data Lake's main cost was not the integration: it was rebuilding every aggregated figure in your own system, then reconciling it with Traackr. The API publishes campaign and creator figures directly, computed the way the Traackr app computes them. See Campaign performance without doing the maths.
2. You fetch only what you need. GET /v1/campaigns?country_iso=FR&start_date_from=2026-01-01 returns this year's French campaigns: not every campaign in the account, and not every post inside them.
3. You control the timing. No export schedule, no waiting for a job to finish. You ask when you want the answer.
2 things you now handle yourself
History. Every Data Lake export is a dated snapshot, kept for a limited time. The API has no equivalent: every call returns the current state, with no "as of" date. If you need point-in-time data (how a campaign's figures evolved, what was reported to a stakeholder in March), store what you fetch, with the time you fetched it, on your own schedule. It is a simple scheduled job, but it was work the Data Lake did for you, so it now needs an owner.
Freshness checks. The API matches the Traackr app within a few hours: data is replicated from the app with a stated window of up to 4 hours, usually much less. A Data Lake export is only as current as its run date. So an export can legitimately show less than the API, and less than the app. See Getting started → How current is the data?.
What looks different on the day you switch
Instagram saves become visible. The Data Lake's Saves column was only ever filled for TikTok. The API fills saves on Instagram posts too, where Instagram reports them. Engagement totals stay the same.
Empty metrics become null. A metric cell that was empty, or showed a dash, is null in the API. It means "not measured", never zero. Spend is the exception: it is 0, never null, when nothing is recorded.
Video views follow the same rule. The Data Lake gave you both Video Views and Video Plays, with a rule to prefer plays. The API applies that rule for you in metrics.video_views; both figures remain in raw_metrics.
Merged creator profiles resolve. In exports, the same post could appear under 2 creator ids across 2 export dates, after a profile merge. The API's uid_aliases on the creator profile links the old id to the current one.
Roster and activated creators are both named. The Data Lake's reports/influencers file held only creators who posted, while projects.csv counted everyone. The API has both on each campaign: stats.creators and stats.activated_creators.
The ids stay the same. Campaign, creator and post ids are identical in both, so you can join your Data Lake history to API data directly.
Some labels become codes. For example, the campaign type Content Creation is SPONSORED_CONTENT in the API. The field mapping gives the details.
No 1% tolerance needed between figures. The Data Lake needed one because its files were written one after another while the account kept moving. In the API, /metrics and the posts reconcile with each other when you read them close together. The stats on each campaign are refreshed on a schedule and can be up to a day behind.
A practical order of work for clients migrating from the Traackr Data Lake
Re-check your KPI definitions. Compare the API's computed totals with what your pipeline produces today, before you switch. Where they differ, the reason is usually a null treated as a zero.
Decide about history. If you need it, build the snapshot job first, not last.
List the fields you actually use, and check each one in the field mapping. A few have no equivalent, and it is better to find out now.
Run both in parallel for one cycle. The Data Lake is a snapshot and the API is current, so they will never match to the unit. Compare shapes and orders of magnitude, and always read the export's run date before calling a gap a defect.
Last reviewed: 30 October 2026.