“We need data from [platform]” is a sentence that sends you down two very different paths depending on what you’re doing with the data. Here’s a short framework for deciding.
The core question: is the data you need a point-in-time or a stream?
Point-in-time = a snapshot. You want every row, once. You’re doing analysis, training a model, backfilling a database, or producing a report.
Stream = live. You want fresh data on demand. You’re powering a search UI, an alerting service, or a product that reflects the source platform in near-real-time.
Point-in-time = dataset. Stream = API.
When a dataset wins
- Research and analytics. “What’s the average price per square foot in Dubai Marina over the last year?” You want historical density that no API rate limit lets you pull efficiently.
- AI / ML training. Models need volume. Datasets give you millions of rows in a single file.
- Internal dashboards. Refresh weekly, show aggregates. You don’t need per-user freshness.
- Bulk enrichment. You have a list of 500,000 records to enrich. An API at 10 req/sec would take 14 hours; a dataset join is 30 seconds.
When an API wins
- Customer-facing search. Your user searched “2-bedroom Marina under 2M AED” — they expect live inventory. The listing that was available last month doesn’t count.
- Alerting. A new listing appeared in the last 15 minutes; your user needs to know.
- Price tracking. You track current prices; historical is nice-to-have.
- On-demand lookups. You need one record at a time, triggered by user action.
When you need both
This is common and often missed. Pattern: a dataset for the cold path (analytics, ML, bulk), an API for the hot path (live product).
Example — a real-estate search product with market-trend insights:
- API: powers the live search. PropertyFinder API returns current listings.
- Dataset: powers the “market trends” tab. PropertyFinder Transactions Dataset has the full history.
Same platform, two products, two integration patterns, one coherent customer experience.
Cost dynamics
Datasets have fixed up-front pricing — buy once, use indefinitely (within licence terms).
APIs scale with usage — more users, more requests, more cost. But you’re paying only for what you consume, and freshness is built in.
For a dataset-heavy product (analytics, BI), datasets are almost always cheaper. For an API-heavy product (live consumer app), APIs are almost always cheaper. The hybrid approach is cheapest when you can cleanly separate cold and hot paths.
How to decide in 90 seconds
Ask:
- Does my product break if the data is 24 hours old? → API
- Will I re-query the same data repeatedly? → Dataset
- Is this for analysis or training? → Dataset
- Is this for a live user experience? → API
- Both cold-path analytics AND hot-path UX? → Both
Where to start
- Browse APIs
- Browse datasets
- Download a free sample before committing to either
Still unsure? Tell us what you’re building — we’ll point you at the right shape.