Cost economics: dollars per account month at 100 vs 1000 vs 10000 scale

most operators who ask me about scaling their account operations are thinking about the wrong problem. they’re focused on whether their antidetect browser can handle more profiles, or whether their proxies are fast enough. they’re not thinking in unit economics. they’re not asking: what does each account actually cost me per month when i factor in every line item, and how does that number change as i 10x my count?

the answer is not linear, and the nonlinearity cuts both ways. some costs drop dramatically with scale. others, particularly the hidden ones, grow faster than your account count. if you don’t model this before you scale, you will discover the math on the other side of a painful month where you spent twice what you expected and earned less than you projected.

this piece is for people who are already running accounts and want to understand what happens to their cost structure as they grow. i’m going to go through each cost category, name actual tools and actual prices where i can, and build out three concrete scale scenarios. i’m not going to be precise to the dollar, because vendor pricing changes and your specific use case matters, but i’m going to be precise enough that you can build a real model.

background and prior art

the unit economics of multi-account operations have never been formally documented in one place. most of what circulates is anecdote: someone in a discord saying “i run 500 accounts for about $800 a month”, which tells you nothing about what’s included, what platform, what proxy type, or what the account acquisition cost was.

the closest academic analog is the economics of bot traffic and automated account activity, which cloudflare has documented from the defender side. what they’ve mapped is that bot operators face an arms race cost structure: detection systems improve, so evasion costs increase, so per-account economics shift. this is background you need to hold in mind. the cost model is not static. what cost you $0.50/account/month in 2022 may cost $1.50 today because the fingerprint surface has expanded and detection has tightened.

the other relevant context is that cloud infrastructure pricing has moved significantly. AWS ec2 on-demand pricing for the instance classes most commonly used in automation work has changed materially over the past three years as spot and reserved pricing dynamics shifted. operators who built models on 2022 spot prices got surprised when those prices moved. the lesson is to model conservatively and recalculate every six months.

the core mechanism

there are five cost buckets in any multi-account operation. i’ll go through each.

bucket 1: proxy costs

this is almost always the largest variable cost, and it’s where the biggest nonlinearities live. you have three proxy types with fundamentally different pricing models:

residential proxies are sold by the gigabyte. at small scale, you’re typically paying retail rates, which at mid-2026 run from about $2.50/GB at the budget end (providers like Smartproxy or IPRoyal on introductory plans) to $8-12/GB at the premium end (Bright Data, Oxylabs). Bright Data’s pricing page gives you a sense of the volume tiers: the per-GB cost drops meaningfully as you commit to larger monthly packages.

the critical variable here is how many GBs each account actually consumes. this varies wildly by platform. a session that’s just logging in and performing a few actions on a lightweight platform might use 10-30MB. a session doing image-heavy social media browsing might use 200-500MB. you need to measure this, not guess.

datacenter proxies are sold per IP per month, typically $0.40-2.00 depending on provider, subnet diversity, and whether they’re shared or dedicated. the unit economics are completely different from residential. your cost doesn’t depend on usage volume, it depends on how many IPs you need and how fast they get burned.

mobile proxies are expensive, typically $50-120 per port per month, but some operations require them because the trust score difference is significant on platforms that care about carrier-level IP reputation.

bucket 2: antidetect browser or browser profile management

for most operations below 1000 accounts, the main tools are Multilogin, AdsPower, Dolphin Anty, and GoLogin. their pricing models differ: Multilogin at the team tier is in the range of $200-300/month for a meaningful number of profiles (check their current pricing at multilogin.com/pricing as these change). AdsPower has a more granular per-profile pricing model starting from a free tier that covers a small number of profiles, with paid plans scaled by profile count.

what matters is not just the headline price but the profile limit. if you’re at 500 accounts and your plan caps at 200 profiles, you’re paying for multiple plan seats or you need to archive and rotate profiles, which has its own operational cost.

above roughly 5000 accounts, the economics of managed antidetect browser services start to break down. at that scale, some operators move to self-hosted solutions, which means either managing a custom Chromium build with fingerprint injection, or running a fleet of real devices. the capex is significant but the ongoing per-account cost can drop substantially.

bucket 3: account creation and maintenance

this is frequently undercounted because operators think of account creation as a one-time cost. it’s not. accounts get banned. accounts age out of usefulness. accounts need periodic re-verification.

the main creation costs are: - SMS verification: prices vary enormously by country and service. a US number on a service like sms-activate or 5sim runs roughly $0.10-0.50 depending on the platform. some platforms require re-verification periodically. - email accounts: if you’re generating these, the cost is near zero per account but there’s automation overhead. if you’re buying aged accounts, prices vary by platform and age. - CAPTCHA solving: services like 2captcha and Anti-Captcha price around $0.50-1.50 per 1000 solves for image captchas. recaptcha v2 and hcaptcha are more expensive, often $1-3 per 1000. this scales directly with how many creation flows you’re running.

ban rate is the multiplier here. if you’re losing 5% of accounts per month, you need to replace them. that replacement cost is per-account creation cost times your ban rate times your total account count. at 10,000 accounts with a 5% monthly ban rate, you’re replacing 500 accounts per month. that’s a non-trivial operation.

bucket 4: compute infrastructure

most automation runs on cloud VMs. for browser automation, the instance sizing matters because headless chrome is memory-hungry. you typically need at least 2GB of RAM per concurrent browser instance, often more.

at small scale, a single $10-20/month VPS handles dozens of sequential sessions just fine. at larger scale, you’re distributing across multiple instances, and you start caring about orchestration, job queues, and failure recovery. that’s both compute cost and engineering cost.

you can see current baseline pricing at AWS, but also look at Hetzner (European-based, often 40-60% cheaper than AWS for raw compute), DigitalOcean, and Vultr. for operations where latency doesn’t matter and you’re doing batch work, Hetzner’s CX-series instances are hard to beat on a cost-per-GB-RAM basis.

bucket 5: engineering and operational labor

this is the most underaccounted cost for operators who came up doing things manually. automation tooling breaks. platforms change their fingerprint detection. proxy providers rotate their pools. the scripts you wrote six months ago stop working.

at 100 accounts, one person can manage this part-time. at 10,000 accounts, you need dedicated technical people or very mature, well-maintained automation infrastructure. the labor cost per account actually increases as you scale if you don’t invest in stability and observability.

worked examples

example 1: 100 accounts, airdrop farming

this is a common entry point. someone is farming crypto airdrops, running 100 wallets with associated social accounts across twitter/x, discord, and one or two protocol-specific platforms.

monthly cost breakdown: - antidetect browser: AdsPower on a mid-tier plan covering 100 profiles, approximately $30-50/month - proxies: residential, because protocol airdrop detection is sophisticated. assume 50MB per account per week of actual activity, so roughly 2GB/month total. at $3/GB that’s $6. but you need IP diversity, so practically you’re buying a minimum package of 5-10GB to have enough IP rotation, making this $15-30/month - infrastructure: single $10/month VPS handles this comfortably - account creation (amortized): if accounts were created 6 months ago and you’re replacing 3% monthly, that’s 3 accounts/month. at $0.20 for SMS plus misc, roughly $1-2/month - total: approximately $55-90/month - per account per month: $0.55-0.90

this is the scale where the economics feel great. low absolute cost, low complexity.

example 2: 1,000 accounts, e-commerce seller management

managing a thousand seller accounts across multiple marketplaces, including account warm-up, listing management, and order handling.

monthly cost breakdown: - antidetect browser: at 1,000 profiles you’re in enterprise territory for most tools. Multilogin’s higher tiers or AdsPower’s business plans. budget $150-250/month. alternatively you might be running GoLogin which has competitive pricing at this scale. - proxies: e-commerce platforms have sophisticated detection. you want residential proxies, and you need dedicated IPs per account for the warm-up period. assume 100MB/account/month for routine operations, so 100GB total. bulk residential pricing at this volume is around $1.50-2.50/GB, so $150-250/month. this is your biggest line item. - infrastructure: need multiple VMs for parallelization. budget $40-80/month. - account creation and recovery: at 1,000 accounts with a 3% monthly churn, replacing 30 accounts/month. SMS verification, email setup, etc. budget $20-40/month. - captcha solving: at this scale captcha costs become real. budget $15-30/month. - labor: this is where it gets serious. 1,000 accounts requires real automation engineering and monitoring. if you’re the only person and this is a side operation, you’re contributing 20-30 hours/month of skilled labor. at any reasonable hourly rate that’s $400-800 worth of time, but operators usually don’t count their own labor. - total (excluding labor): approximately $375-650/month - per account per month: $0.38-0.65

note how per-account cost actually dropped compared to 100 accounts. this is the bulk pricing effect on proxies and the fixed-cost dilution on browser tooling.

if you’re curious how proxy selection at this scale gets analyzed, the antidetect browser comparison guides at antidetectreview.org/blog/ are worth reading alongside this cost framework.

example 3: 10,000 accounts, traffic and engagement operation

this is a serious operation. 10,000 accounts running coordinated engagement across social platforms.

monthly cost breakdown: - antidetect browser: at 10,000 profiles, managed SaaS tools are either prohibitively expensive or they cap out. you’re likely running self-hosted browser infrastructure. initial capex of $2,000-10,000 to build the system, then ongoing maintenance. amortize capex at $300-500/month, plus engineering labor for maintenance. - proxies: this is where you need serious proxy infrastructure. 10,000 accounts at even 30MB/month each is 300GB. at enterprise residential pricing you might negotiate $0.80-1.20/GB, making this $240-360/month. more likely you’re mixing proxy types: a pool of 2,000-3,000 datacenter IPs at $0.50/month each ($1,000-1,500/month) supplemented by residential for sensitive operations. - infrastructure: multiple VMs, orchestration layer, monitoring, a job queue system. budget $200-400/month. - account creation and recovery: at 3% monthly churn that’s 300 accounts/month to replace. with SMS, email, and captcha costs, budget $150-300/month. - captcha solving: scales directly. 300 creations plus ongoing verification flows. $60-120/month. - labor: at this scale you have at least one full-time technical person or equivalent contractor. say $3,000-5,000/month for skilled labor. - total: approximately $1,950-2,680 excluding labor, $4,950-7,680 including one full-time technical resource - per account per month (excluding labor): $0.20-0.27 - per account per month (including labor): $0.50-0.77

the per-account cost excluding labor drops significantly. but the labor costs at this scale are real and large in absolute terms.

for context on proxy cost structures at this kind of volume, proxyscraping.org/blog/ covers provider selection and cost optimization in depth.

edge cases and failure modes

failure mode 1: ban cascades

you build a cost model assuming 3% monthly ban rate. then you make a configuration mistake, the platform rolls out a new detection signal, or your proxy pool gets flagged, and you lose 40% of your accounts in 72 hours. this is not a theoretical scenario, it happens regularly.

counter-strategy: model your maximum loss scenario, not your average scenario. what does it cost to replace 30% of accounts in a week? do you have the capital buffer to absorb that? if not, you’re one bad detection event away from being underwater.

failure mode 2: proxy cost explosions from uncontrolled usage

residential proxies billed by the gigabyte will bankrupt you if your automation code has a bug that causes runaway requests. i’ve seen operators hit 10x their expected GB usage in a single night because a retry loop went infinite.

counter-strategy: set hard GB caps on your proxy account at the provider level, not just in your code. every major residential provider has a cap you can set. use it. also, measure actual per-session bandwidth before scaling, not after.

failure mode 3: the hidden platform-specific cost

some platforms require phone-verified accounts, and the phone numbers need to be valid for the lifetime of the account for re-verification. that means you either need to hold numbers (recurring cost on services that charge monthly for number retention) or accept a higher account replacement rate. operators who don’t account for this discover it when their entire account base suddenly requires re-verification.

failure mode 4: tooling version lock

you build your automation stack around a specific version of a headless browser or an antidetect tool. the platform updates its fingerprinting. your tool vendor releases an update that breaks your automation scripts. now you have a choice: emergency engineering work, or a period of degraded operations.

counter-strategy: build abstraction layers in your automation code so that swapping the underlying browser tool doesn’t require rewriting everything. this is engineering overhead upfront but it reduces the blast radius of vendor changes. the airdrop farming operations at airdropfarming.org/blog/ have written about this in the context of rapid platform changes.

failure mode 5: labor cost compression at scale

operators scaling from 100 to 10,000 accounts often assume labor costs scale sublinearly. sometimes they do. but if your automation is not mature, or if your monitoring is weak, labor costs can scale faster than account count. you end up hiring or spending more time firefighting as you grow.

counter-strategy: invest in observability before you need it. you want per-account success rate metrics, proxy health dashboards, and automated alerting. the cost of building this at 500 accounts is much lower than rebuilding at 5,000 accounts when things are on fire.

what we learned in production

running operations at different scales, the pattern i keep seeing is that operators optimize the wrong costs. they spend hours trying to find cheaper proxies when the real cost driver is account replacement rate. a 2% ban rate versus a 5% ban rate matters far more than the difference between $2/GB and $2.50/GB residential proxies, both in direct replacement cost and in the labor cost of managing higher churn.

the other pattern: fixed costs like antidetect browser subscriptions feel expensive at small scale and cheap at large scale, which creates an incentive to scale to dilute them. but fixed costs are not a good reason to scale. scale when your unit economics at current scale are already positive and your marginal account is profitable. scaling to dilute fixed costs while your per-account revenue is unclear is how operations blow up.

there’s also a real cost to the time you spend on security posture and maintenance of browser profiles. i’ve seen a per-account “profile hygiene” cost that operators don’t count: the time to warm accounts, age them, maintain consistent activity patterns, update fingerprints when tool vendors push breaking changes. if you price your time honestly, this often doubles the apparent per-account cost, especially at scale.

the practical implication for planning: take whatever per-account cost estimate you have, double it to account for untracked labor and edge cases, and model against your revenue. if it still works at that inflated cost, you probably have a real operation. if it only works in the best case, you’re one bad month away from learning an expensive lesson.

for more context on how the tooling landscape for browse-based operations actually compares across providers, see our antidetect browser comparison guide and our overview of residential proxy providers for multi-account work. for a more foundational take on building the infrastructure layer, see setting up your first account farm.

references and further reading

1. Cloudflare: what is a bot? - cloudflare’s documentation on automated traffic, including detection mechanisms that directly affect cost structure for operators.

2. AWS ec2 on-demand pricing - authoritative current pricing for the compute infrastructure layer. check hetzner and vultr for comparison.

3. Bright Data pricing - one of the more transparent pricing pages for enterprise-tier residential proxies, useful as a benchmark for volume pricing tiers.

4. Multilogin pricing - current pricing for one of the primary antidetect browser platforms, illustrates how per-profile pricing scales across tiers.

5. FTC guidelines on disclosure - relevant if your account operations include any affiliate or endorsement activity, this is not legal advice.

Written by Xavier Fok

disclosure: this article may contain affiliate links. if you buy through them we may earn a commission at no extra cost to you. verdicts are independent of payouts. last reviewed by Xavier Fok on 2026-05-19.