[ USE CASE / MARKET RESEARCH ]

See the market as it really appears

Collect accurate, location-specific market data across retailers, marketplaces, search engines and public websites — without blocks, geographic bias or incomplete samples.

$0.99/ 1 GBRESIDENTIAL · PREPAID · $0.49 / GB AT 1 TB

Buy the smallest amount of traffic, run one market as a pilot, then widen the sample. Plans are valid for 30 days from activation.

Start collecting data Explore proxy plans

137 COUNTRIES
CITY AND ASN TARGETING
STICKY RESEARCH SESSIONS
REPEATABLE SAMPLING METHOD

LIVE MARKET PULSE run mr-1842 COLLECTING T+00:00 0 ACTIVE

MARKET RESEARCH SIGNAL EXIT SESSION

ILLUSTRATIVE LIVE SIMULATION — NO CUSTOMER TRAFFIC SHOWN

REQUESTS COMPLETED0

ACTIVE MARKETS0

OBSERVATIONS0

TRANSFERRED0MB

MEDIAN LATENCY0ms

SAMPLE COVERAGE0%

Illustrative sample data. Every market, source name, value, status and figure in this console is generated on the page as an example of what a research run can return — not measured results, not customer traffic and not a coverage promise.

[ 01 / THE RESEARCH GAP ]

One market. Many different realities.

Websites decide what to show from the outside in: the location the request came from, the reputation of the address, the history of the session and the shape of the access pattern. Read a German catalogue from a hosting range in Virginia and you are not looking at the German market — you are looking at what that site serves to an unrecognised visitor.

ONE SKU · ONE URL · 4 MARKETS · SAME MINUTE3 WAYS THE SAMPLE BENDS

01GEOGRAPHIC BIAS

The page changes with the address

Prices, inventory, rankings and offers change between regions — often on the same URL, decided by the address the request came from.

SAME SKU · 3 DIFFERENT PRICESΔ 11%

02ACCESS FRICTION

The hardest sources drop out

Rate limits, CAPTCHAs and outright blocks remove exactly the sources that matter most, so the sample quietly stops representing the market.

DISCARDED FROM A 25-SOURCE SAMPLE11%

03DYNAMIC EXPERIENCES

The value arrives after the render

JavaScript rendering, personalisation and session history mean two collections of the same page can legitimately differ.

PRICE NODE PRESENT ONLY AFTER RENDER0 → 1

[ THE POINT ]Research decisions are only as reliable as the market view behind them.

[ 02 / WHAT YOU CAN MEASURE ]

Six questions a local sample can answer

Each of these is the same mechanic — read a public page as a local visitor, repeat it on a schedule, keep the conditions attached. What changes is the question you ask of the result, and the shape the answer takes.

M01BAR COMPARISON

Competitive landscape

Track competitors, categories, positioning and distribution across markets.

COMPETITORS FOUND PER MARKET

US31

DE23

JP17

M02INDEXED PRICE

Pricing & assortment

Compare prices, promotions, availability, variants and product catalogues.

SAME SKU, INDEXED TO US PRICE

US100

UK88

DE104

M03RANK DISTRIBUTION

Search & share of voice

Measure brand visibility across search results, marketplaces and directories.

BRAND APPEARANCES, POSITIONS 1–10

M04SENTIMENT SHARE

Reviews & sentiment

Collect public reviews, ratings and recurring customer themes.

RECURRING THEMES BY SHARE

DELIVERY 47%SIZING 31%PRICE 22%

M05CHANNEL COVERAGE

Market entry research

Validate demand, competition and channel coverage before entering a region.

CHANNEL COVERAGE OBSERVED

RETAIL82%

MARKETPL64%

D2C35%

M06TREND LINE

Trend monitoring

Repeat the same collection method over time to identify meaningful changes.

PROMOTION DEPTH, LAST 4 WAVES

W1 18%W2 24%W3 39%W4 46%

Every panel above is an illustrative shape, not a measured result — the sample, the market and the wave decide the real numbers.

[ 03 / RESEARCH WORKFLOW ]

From research question to evidence

One continuous pipeline, and the fourth stage is the one that makes the work defensible: an observation without its geography, timestamp and method is a number nobody can check.

01DEFINE

Define the sample

Choose sources, markets, categories and collection frequency — and write them down, because that list is the study.

OUTPUT · SOURCE LIST

02COLLECT

Collect locally

Send every request through a residential exit in the relevant market, with a session stable enough to finish the page.

OUTPUT · RAW RESPONSES

03RENDER

Render when required

Use browser-aligned TLS and a real render for JavaScript-heavy or protected sources; skip it where plain HTML carries the data.

OUTPUT · COMPLETE PAGES

04NORMALIZE

Normalize the results

Keep geography, timestamp, source and collection method attached to every observation. This stage is what makes the study defensible six months later.

THE DEFENSIBLE STEP

05EXPORT

Export the evidence

Send results into your database, spreadsheet or analytics workflow — raw response kept alongside the parsed fields.

OUTPUT · EVIDENCE ROWS

PLATFORM CAPABILITIES USED

COUNTRY, CITY AND ASN TARGETING
ROTATING AND STICKY SESSIONS
TLS / BROWSER-ALIGNED REQUESTS
AUTOMATIC RETRY HANDLING
API-FRIENDLY INTEGRATION
REPEATABLE GEOGRAPHIC SAMPLING

Gateway: gw.ludaxproxy.com:7777 (Residential · HTTP and HTTPS)
Parsing: Yours. LUDAX supplies the location, the session and the network path; extraction and analysis stay in your tooling.

Parameter reference

[ 04 / RESEARCH SCENARIOS ]

Three shapes of research sample

The mechanics do not change between these; the configuration does. Select a study blueprint to see how the sample, the session behaviour and the monthly traffic move with it.

[ 06 / FIELD MANUAL & RESPONSIBLE USE ]

Read this before you run a wave

HARD LIMIT

Public, published information only.

Research value comes from material any visitor in that market can load: catalogues, pricing pages, public listings, published reviews. Not accounts, not internal dashboards, not systems behind an access control, and no personal data without a lawful basis. Respect each source's terms and robots directives, and keep the work inside the acceptable use policy.

Collection is read-only: sample at a rate the source can absorb. A study that degrades the site it measures is not research.

One observation is not a finding

Prices rotate, stock moves and rankings are personalised. Sample the same source several times per market before you treat a difference as structural, and keep every raw response so the sample can be re-read.

DO repeat the sample · keep raw HTML · report ranges, not single readings

Record the collection conditions

A number without its geography, timestamp and method cannot be defended six months later. Country, city, ASN fallback, session, device profile and render mode all belong in the row.

DO store -country / -city / -session and the exact run time with each observation

Render only where you must

A rendered catalogue page with imagery can be twenty times an HTML fetch, and traffic is metered to the kilobyte. Strip assets when the text and availability are the data.

DO size the month with the estimator above before a wide scan

Hold the session for multi-step sources

Search, filter and paginate inside one sticky session, or the source treats each step as a new visitor and the sample stops being coherent.

DO -session-mr14-time-10 · retry on a fresh token rather than reusing a failed one

No guarantees. What a source returns depends on the source: its geo logic, its bot defence and its own personalisation. No proxy network can promise a challenge-free collection, and no sample can prove a market-wide fact on its own.

collect_market.py

import requests, datetime

USER, PASS = "YOUR_USERNAME", "YOUR_PASSWORD"

def proxy(country, city=None, session=None):
    pw = f"{PASS}-country-{country}"
    if city:    pw += f"-city-{city}"
    if session: pw += f"-session-{session}-time-10"
    url = f"http://{USER}:{pw}@gw.ludaxproxy.com:7777"
    return {"http": url, "https": url}

# same sample, one sticky session per market, conditions recorded
SAMPLE = [("us", "new-york"), ("gb", "london"), ("de", "berlin"), ("jp", "tokyo")]

for cc, city in SAMPLE:
    p = proxy(cc, city=city, session=f"mr-{cc}")
    r = requests.get("https://example-retailer.com/c/outdoor", proxies=p, timeout=30)
    region = requests.get("https://api.ludaxproxy.com/whoami", proxies=p, timeout=15).json()

    # the row is only comparable if it carries how it was collected
    row = {"market": cc, "city": region.get("city"), "asn": region.get("asn"),
           "captured_at": datetime.datetime.utcnow().isoformat() + "Z",
           "method": "html", "bytes": len(r.content), "html": r.text}

Full parameter reference in the documentation.

[ FAQ ]

Market research FAQ

01Which product fits market research?

For open, country-level collection standard Residential is enough. Sources that render prices or listings in JavaScript, or that inspect the connection, are the case for Residential TLS; persistently challenged sources are the case for the screened premium pool.

02How do I keep two waves comparable?

Freeze the sampling method, not just the URL list: same country and city parameters, same session behaviour, same device profile, same time of day. Record all four with every observation or the second wave is a different study.

03Is city or ASN targeting available everywhere?

City targeting is available in most markets, ASN targeting in fewer. Where a city has no capacity the request resolves at country level — record that fallback with the sample rather than discarding it.

04How much traffic does a study need?

A focused study across two markets is usually under a gigabyte. A multi-market scan of 25 sources across 8 markets, sampled four times a month, lands near 30 GB. Size yours with the estimator above before you buy.

05Do you return structured data or reports?

No. LUDAX provides the network path, the geography and the session control; parsing, normalisation and analysis stay in your own tooling.

[ START HERE ]

Sample one market. Then the whole set.

Buy the smallest amount of traffic, collect one market's sources as a pilot, and check the sample holds up before you widen it.

$0.99/ 1 GB · $0.49 / GB at 1 TB

Start with 1 GB — $0.99 See full pricing

[ RELATED USE CASES ]