Proxies for Reverb read a music-gear marketplace whose value is a price spread, from behind a two-layer door. On 24 August 2026 we sent reverb.com a plain request from a home connection and it loaded, HTTP 200, with F5 Shape telemetry and currency-and-region cookies set from the exit. The identical request from a datacenter server got Cloudflare's challenge. So the door is Cloudflare, trusting the residential IP and challenging the datacenter one, with F5 Shape scoring the session behind it.
Reverb is the largest marketplace for new, used and vintage musical instruments, and its data is worth reading for the same reason any marketplace is: a single model carries many listings at many prices and conditions, plus Reverb's own Price Guide of estimated market values. That makes the proxy job a per-seller, per-condition read behind Cloudflare and Shape, in the same family as Grailed and TCGplayer.
What proxies work best for Reverb?
US rotating residential proxies behind a real browser, one exit per session because Shape scores behavior. A static ISP proxy for a logged-in account. Datacenter proxies get challenged on contact.
Cloudflare scores the IP, Shape scores the session
Cloudflare distrusted our datacenter address and served the challenge, while the residential address loaded; it reads IP reputation and the TLS fingerprint, so a residential IP behind a real browser is what passes (how to scrape past Cloudflare, what is JA3/JA4 fingerprinting). F5 Shape then scores the session over time on movement, timing and JavaScript execution (how to handle F5 Shape), so one consistent exit per session beats rotation. The combination is a residential IP for reputation, a real browser for the fingerprint and Shape, and consistency across the session.
The price is a per-seller spread plus a guide
Reverb's data has two useful layers. The listings are a spread: thousands of shops and individuals sell the same model new, used and vintage, at prices that vary by condition and seller, so a price project captures the listing-level spread and the condition, not one blended number. And the Reverb Price Guide publishes estimated market values per model, a second reference worth recording alongside the live listings. Currency and shipping region localize by exit (the site set reverb_user_currency and reverb_user_country_code from our IP), so cross-country comparison needs an exit in each market (geo-testing with proxies confirms the exit).
The proxy solves
IP reputation
residential passes, datacenter challenged
Market and currency
an exit per market
Session isolation
one address per account
Only the browser solves
The Shape behavior score
movement, timing, JS
TLS fingerprint
read at the Cloudflare gate
Consistency per session
no mid-session IP swap
Which proxy type fits which job
| Reverb job | Proxy type | Why |
|---|---|---|
| Per-seller, per-condition price capture | US rotating residential, one exit per session | Cloudflare plus Shape score IP and behavior |
| Price Guide value collection | Rotating residential | A second reference per model |
| Cross-country price comparison | Residential in that market | Currency and shipping localize |
| Seller and listing tracking | Sticky residential per session | Following a seller over time |
| Seller or buyer account | Static ISP | The session must keep one address |
| Parser tests against saved HTML | Residential (datacenter is challenged) | Datacenter draws the challenge |
Rotating residential IPs pass Cloudflare and start clean with Shape, and ISP proxies hold one address for an account. The two are compared in rotating vs static residential proxies.
Setup
Drive a real browser and hold one residential exit per session, because Shape scores behavior (proxies for Playwright). Capture the listing spread and condition per model, record the Price Guide value alongside, and pin the market with the currency for cross-country reads. For accounts, use a static ISP address. Pace like a shopper.
Sizing
Size by sessions, because Shape scores the session, and by market. Decide how many listings one shopper-paced session reads before the score tips, and add exits, with a pool per market you price. Budget browser-level bandwidth (headless browser proxy bandwidth cost). Accounts are one static ISP address each, and our pricing is pay-as-you-go with a balance that does not expire.
Free versus paid for Reverb
A free datacenter IP draws Cloudflare's challenge and Shape scores it down, so free is for prototyping; our free proxy list and proxy checker cover that. Monitoring the price spread needs residential from $0.44/GB pay-as-you-go with no KYC behind a real browser. The line is in datacenter vs residential proxies.
Staying unblocked
- Do not rotate mid-session. Against Shape, an IP swap inside a session is a bot tell.
- Do not retry the challenge from the same datacenter address. Use a residential exit and a browser.
- Capture the spread and condition, plus the guide value. One blended price hides the marketplace signal.
- Pin the market with the currency. A price without its currency and region is incomplete.
- Watch the challenge rate. A rising share means slow down or widen the pool. The full list is in avoiding IP bans while scraping.
What a proxy does not do here
A proxy gives a browser a clean, market-correct address and keeps accounts apart. It does not make a headless client behave like a person for Shape, and it does not turn the listing spread into one number, which is the whole point of the data. Scraping reverb.com runs against its terms of use, a risk that stays with you regardless of the IPs. What a residential exit does is load the marketplace a datacenter client cannot and keep a well-behaved session trusted, after which reading the per-seller spread and the guide value is the work. Test with the free proxy list, then monitor with residential from $0.44/GB, one exit per session.
Sources
- Reverb, Price Guide: the estimated market values per model that complement the live listing spread.
- Cloudflare, bot management: the layer that challenged our datacenter IP.
- F5, Shape / Distributed Cloud Bot Defense: the session-scoring layer present in the page.
- HProxy test on 24 August 2026: identical plain GET to reverb.com from a residential IP (HTTP 200, Shape telemetry and currency/region cookies) and a datacenter IP (Cloudflare challenge).