The Exit Audit: Twelve Questions That Reveal What Your Scraping Vendor Really Costs to Leave

Nobody prices the exit when they sign. Then a budget crunch, a better rate, or a compliance mandate arrives, and the switching cost turns out to be the most expensive line in the vendor’s favor. This is an exit audit — twelve due-diligence questions to ask any scraping or proxy vendor before signing (asked here against Bright Data and Thordata, both of which are excellent products; the audit isn’t about quality, it’s about lock-in), plus the answers that should make you comfortable either way.
The best time to run an exit audit is before you sign. The second-best time is now.
Why Exit Costs Are the Hidden Price
Vendor comparisons fixate on per-GB rates, but the total cost of a scraping vendor is entry + usage + exit, and exit is where structure bites: data-format lock-in, SDK coupling, contract minimums, migration engineering, and the institutional knowledge wrapped in one account manager’s Slack thread. Two vendors with identical rates can differ by an order of magnitude in what it costs to leave.
Bright Data — one of the industry’s most complete platforms, with an enormous IP network, mature unlocker products, and enterprise compliance packaging — prices that completeness partly through structure: monthly plan commitments (published residential rates from $5–7/GB on $499–$1,999/month plans, $8/GB pay-as-you-go), an enterprise sales motion, and ecosystem depth that teams organically build around. Thordata prices for the opposite posture: a published self-serve slider ($2.00/GB down to $0.65/GB at volume, 1 GB minimum, no plan buckets), standard proxy endpoints, and per-result/per-response API pricing (~$0.50–$1.00 per 1,000 results; ~$0.70–$1.20 per 1,000 SERP responses). Neither posture is wrong. The audit questions are how you find out which one your organization is actually buying.
The Twelve Questions
Contract and commitment (the money exits)
- What is the minimum monthly commitment, and what happens to unused volume? Plan-bucket pricing (both vendors’ enterprise tiers, and the industry norm) sells predictability and buys stickiness; slider pricing has no minimum to strand. Know which you’re holding.
- What does termination mid-term cost? Ask for the clause, not the summary. The difference between 30-day notice and contract-length liability is an exit cost you’re choosing at signing.
- Are negotiated rates contingent on volume? Custom rates (common above 1 TB/month on enterprise plans) often carry volume floors — check whether a usage dip reprices you retroactively or prospectively.
Data and integration (the engineering exits)
- Is our data exportable in a standard format, today, in bulk? If your collection history lives in a vendor-proprietary schema, the exit includes a data-migration project. Standard JSON/CSV/XLSX outputs and raw proxy responses make your warehouse vendor-agnostic.
- How much of our code touches vendor SDKs versus standard protocols? Standard proxy endpoints (username/password auth,
gateway:port) mean a one-line migration; deep SDK coupling means a sprint per service. Both vendors here offer standard endpoints — the audit question is how much *your team* uses them versus the proprietary layers. - Do scheduled jobs, session configurations, and dashboards have export equivalents? The invisible exit cost: months of tuned scraping configs that don’t transfer. Ask what’s portable before you build your operation inside any dashboard.
Operational (the knowledge exits)
- Which runbooks reference vendor-specific behavior? Every team accumulates tribal workarounds (“retry twice on Tuesdays, the pool is slow”). An exit without a runbook audit exports the workarounds as broken code.
- What’s the parallel-run path? Any vendor that makes running 10% of traffic through a competitor awkward (by contract or by technical coupling) is charging you an exit premium monthly. Standard endpoints make parallel-running a configuration exercise.
- Who owns the relationship knowledge? Enterprise motions concentrate context in account managers — excellent while you’re a customer, and part of the exit cost when you’re not. Self-serve models keep the knowledge in your docs and dashboards.
Strategic (the future exits)
- If our volume 10×’d, does our pricing improve automatically or by negotiation? Slider and per-result pricing improve mechanically; plan tiers improve by a sales conversation. One of these scales with you without a renegotiation event.
- If our volume collapsed 10×, could we scale down tomorrow? The symmetric question — bursty and seasonal workloads get hurt by minimums in both directions.
- Could we run a second vendor alongside this one without anyone’s permission? Multi-vendor is the practical hedge in production scraping; a contract or technical structure that discourages it is a lock-in artifact worth pricing.
Scoring the Two Postures (Not the Two Products)
Run the twelve against both vendors and the scorecard sorts by posture, not quality:
| Question cluster | Bright Data posture | Thordata posture |
|---|---|---|
| Contract exits (1–3) | Enterprise plans, negotiated rates, sales motion | No minimum beyond 1 GB, published slider, no repricing events |
| Engineering exits (4–6) | Standard protocols available; ecosystem depth encourages coupling | Standard endpoints and formats by default; per-result APIs portable |
| Knowledge exits (7–9) | Concentrated in account management | Concentrated in your configs and docs |
| Future exits (10–12) | Improvement via negotiation (and it’s real — enterprise buyers get real terms) | Improvement via mechanics |
The honest reading: Bright Data’s structure is the *point* for organizations that need enterprise agreements, DPAs, SLAs, and named support — those artifacts are the product, and the lock-in they create is the cost of having them. Thordata’s structure is the point for self-serve teams that want the exit door to stay unlocked. If you need the enterprise package, pay for it knowingly; if you don’t, notice that you’re holding it.
FAQ
Isn’t asking exit questions at signing adversarial?
Vendors respect it — procurement teams that ask exit questions are easier to serve, because they’ve thought about the relationship’s shape. Any vendor uncomfortable with the twelve is answering question zero for you.
We’re mid-contract. Is the audit still useful?
More useful — it builds your exit runbook before you need it: export your data monthly, keep a standard-protocol shim between your code and any SDK, document the vendor-specific workarounds. Exit costs are partially a choice you keep making (or not making) every month you stay. And the cheapest place to start practicing dual-vendor posture is the search workload: SERP monitoring on a mirror slice of your keywords, free trial credits first, structured output from day one.
What’s the single highest-leverage exit question?
Number 8, the parallel-run path. A team that can mirror 10% of traffic through a second vendor at any time — standard endpoints make this an afternoon’s work — holds every negotiation, every outage, and every budget review from a position of strength. Put the second vendor on the SERP workload first: rank monitoring is the easiest workload to dual-run, priced per structured response at ~$0.70–$1.20 per 1,000, with free trial credits — and if the mirror shows parity on your keywords, you’ve converted an abstract exit option into a tested one.
Does an exit audit mean we’re planning to leave?
It means the opposite: you’re planning to stay *deliberately*. Teams that never audit their exit end up staying by default, paying whatever the renewal says; teams that audit stay by choice, at the rate the market supports. Run the twelve before the next signature — or this afternoon, with a dual-run pilot as the practical exercise — and the next vendor conversation you have will be the first one where all the costs are on the table, including the ones nobody quotes.