A European ticketing platform is not one site. It is eleven country sites with different layouts, languages, currencies, tax treatments and even different ideas of what a seat is — and a client trying to compare markets needs all of that arriving as one clean table.
The source is named because it is public; the client is not. Every figure on this page is a measured production number the client agreed to publish. Have a source of your own? Send it over, whatever its size — you get an answer within 24 hours.
event records to date
a year
measured coverage
events tracked
Not more rows. Every one of these projects started with somebody who already had data and could not use it for the decision in front of them.
The problem
The client operates in several European markets and wanted to see all of them the way a single analyst would. What they had was eleven separate exports in eleven shapes, reconciled by hand each month by somebody who was the only person who understood the differences — and who was, inevitably, about to leave.
Fixed on day one
Every one of these is measured continuously and reported on the same dashboard the client watches. A commitment nobody measures is a sentence in a proposal.
None of these is solved by better headers or a bigger proxy pool. Each needs a different piece of engineering, and working out which one you are actually facing is most of the job.
Market sites share a brand and very little else. Layouts, identifiers, pagination and even the meaning of a category differ, so eleven collectors is the obvious answer and eleven collectors is a maintenance disaster.
What we doOne collector with a per-market adapter layer: the traversal, the reconciliation and the schema are shared, and only the market-specific reading is separate. A new market is an adapter, not a project.
Seat tiers, price categories and event genres carry local names that do not map one to one. Naive mapping produces comparisons that look right and are not.
What we doA maintained taxonomy with explicit per-market mappings, reviewed with the client, and an unmapped-value alert so a new local category surfaces as a question rather than silently becoming 'other'.
Six currencies, different VAT treatment and fees that are sometimes included in the displayed price and sometimes not. Averaging across that produces numbers nobody can act on.
What we doPrices are stored in local currency with the observed rate and the tax treatment as attributes; the normalised base-currency figure is derived, never the only thing kept.
A layout change in one market is invisible in aggregate metrics — total volume barely moves — and can go unnoticed for weeks.
What we doCoverage and field-fill rates are monitored per market rather than in aggregate, so a single market drifting triggers an alert on its own.
The architecture is not interesting; every extraction system has a queue, a fetcher and a parser. These are the decisions that made this one work where the last one did not.
The expensive parts — traversal, scheduling, reconciliation, schema — are written once. Only the market-specific reading is per market, which is what keeps eleven markets from costing eleven times as much.
Local categories map to a shared taxonomy through a reviewed table, not a heuristic. Anything unmapped raises an alert instead of being quietly bucketed.
Local currency, local tax treatment and the observed exchange rate are all retained. The normalised figure is computed on top and can be recomputed when the client changes their mind about the method.
Every quality metric exists per market. Aggregate monitoring is how a single broken market hides for a month behind ten healthy ones.
Adding a twelfth market is an adapter, a taxonomy review and a day of verification. That was the actual requirement behind the brief, even though it was not in it.
Production figures, not a benchmark run. Coverage is recounted daily against an independent sample of the live source rather than asserted, which is why the numbers are not round.
Daily profile
Twelve million records a day is 360 million a month, 4.4 billion a year and 15 billion since 2022, from 1.4 million requests — sixteen a second — because an event page carries its whole price-category table. A dozen unmapped values a week is the number that keeps the taxonomy honest: a system that never raises one is not detecting change, it is hiding it.
Headline
These four sit in the support agreement. When one of them drifts outside its band, we are alerted within fifteen minutes and fixing it is routine work under the monthly arrangement, not a change request.
The columns are the client's own numbers from before the rebuild and the measured ones from production today. The left column is the part most vendors would rather not put on a page.
| Metric | Before | Today |
|---|---|---|
Markets in one view |
None — eleven exports | All eleven, one schema |
Monthly reconciliation |
Two days of manual work | None |
Cross-market comparison |
Approximate, by hand | Direct, with tax and currency kept |
Adding a market |
A project | A day |
The person who used to reconcile the eleven exports by hand now works on the product. That was the outcome the client actually wanted.
One case file per source, each opening the same way: what the client needed, what the source put in the way, how the pipeline is built and what the numbers are today.
The whole UK market, under a search ceiling
Read the case →
Three countries, 12M properties, one model
Read the case →
The market, and the 1.9M people selling it
Read the case →
7.4M listings, 210M calendar-nights a day
Read the case →
46M itineraries priced a day, decomposed
Read the case →
Start here
Send the source, the fields you need and roughly how often. You get a straight answer within 24 hours: whether it can be done, what makes it hard, what coverage is achievable and roughly what it costs to build and to run. Whatever the size.
Direct
NDA before technical detail, as always. If it isn't our kind of work, we say so in the first reply.