If you want to know where a page ranks — for a keyword, in a country, on a given day — you need to see a search results page the way a searcher in that place would see it. Doing that once is trivial. Doing it for five hundred keywords across three locations every morning is a different problem entirely, and it is the problem rank tracking APIs exist to solve.
We are Geonode, and we sell proxies — which makes this an awkward article to write honestly, so let us get it out of the way in the first paragraph. For most people reading this, a rank tracking API is the thing you should buy instead of proxies from us. If your requirement is "give me the top 100 results for these keywords, parsed, daily", a managed SERP API does the proxy rotation, the CAPTCHA handling, the parsing and the parser maintenance for a price that is very hard to beat by assembling the pieces yourself. Proxies become the right purchase further down the article, in a specific set of cases we will name precisely.
The second thing worth saying up front is that pricing in this category is close to unreadable. One provider quotes a monthly subscription with a search allowance. Another quotes dollars per thousand requests. A third quotes credits, where a credit is not always one search. Run a realistic workload through the actual numbers — we do this later with 500 keywords across 3 locations, checked daily — and the monthly cost ranges from about $27 to about $725 for the same data. That is not a rounding error, and it is invisible if you compare the front pages.
All prices below were taken from providers' own pricing pages at the time of writing. They change often, sometimes monthly. Treat every figure here as a starting point for your own check rather than a quote, and follow the links before you commit to anything.
What a Rank Tracking API Actually Is
At its simplest: you send a keyword, a location and a search engine to an HTTP endpoint. You get back structured JSON describing what appeared on that results page — organic results with their positions and URLs, plus whatever else was on the page.
That last part matters more than it used to. A modern results page is not ten blue links. It has ads, featured snippets, People Also Ask boxes, local packs, shopping carousels, video blocks, and increasingly AI-generated summaries sitting above everything else. "Position 3" means something very different on a page where positions 1 and 2 are pushed below a full-screen answer box. A good API tells you what else was there; a poor one hands you a ranked list and lets you draw the wrong conclusion.
The Two Products Sold Under One Name
Search for "rank tracking API" and you will get two categories that solve different problems. Conflating them is the most common way people end up with the wrong tool.
Raw SERP APIs. You ask, they fetch a live results page, they parse it, you get JSON. Nothing is stored. There is no notion of "your keywords" or "your project" — every call is independent. SerpApi, DataForSEO, Bright Data, Oxylabs, ValueSERP and Serper all sit here. You are buying a fetch-and-parse service, and you build the tracking layer on top.
Rank trackers with an API attached. These are products with a database, a dashboard and a schedule. You register keywords, they check them on a cadence, they store history, they compute changes and visibility scores, and the API is a way to read that data out. AccuRanker, SE Ranking, Semrush and Ahrefs are in this group.
The distinction decides your architecture. A raw SERP API means you own the storage, the scheduling and the history — more work, far more control, usually much cheaper per data point, and the only option if you want SERP features, competitor pages or anything the tracker's schema does not model. A tracker-with-API means someone else owns the history and you get change alerts and reporting for free, at a price per keyword rather than per request.
A rough rule: if you are building a product, a client dashboard or an internal system that needs raw results, buy a SERP API. If you need a team to look at rankings in a browser and occasionally pull numbers into a spreadsheet, buy a rank tracker and use its API as an export.
The rest of this article is mostly about the first category, because that is where the interesting cost decisions live.
Why Rank Tracking Needs Proxies at All
This is worth understanding even if you never touch a proxy, because it explains what you are paying for.
Search engines do not want to be scraped. Automated queries cost them money, return nothing, and hand competitors their index. So they invest heavily in telling automated traffic apart from human traffic — and the signals they use are not subtle.
Address origin. Requests from data centre IP ranges are visibly not ordinary users. Those ranges are publicly registered and easy to enumerate.
Rate and rhythm. A human does not issue two hundred queries a minute at perfectly regular intervals. Timing patterns are among the easiest things to spot.
Browser fingerprint. Header order, TLS handshake characteristics, JavaScript execution behaviour. A default HTTP client looks nothing like a browser at the transport layer, regardless of what its user-agent string claims.
Session behaviour. Cookies, referrers, whether anything on the page ever gets clicked.
The usual outcome of failing these checks is not a clean error. It is a CAPTCHA, or a page that renders differently, or results quietly reordered. The dangerous failure is the silent one — you get a 200 response and a plausible-looking page, and your data is subtly wrong. That is worse than an outage, because nothing alerts you.
Location Is the Other Half
Search results are localised, and not only by country. A query for a service business returns different results in Manchester than in London, and different again in Leeds. If you need to know how a page ranks for users in a specific city, the request has to plausibly originate near that city — or use the engine's own location parameters, which approximate it but do not always match what a real user there sees.
This is the genuine reason proxies enter the picture. Geographic accuracy is not a bonus feature of rank tracking; for local businesses it is the entire point.
A managed SERP API absorbs all of this. You send a keyword and a location string; behind the scenes someone else is rotating addresses, managing sessions, solving challenges and repairing parsers when the results page changes shape. That maintenance burden, not the bandwidth, is what you are actually buying.
What Actually Differentiates These APIs
Six things, in roughly the order they will bite you.
Success rate on your queries. Not the advertised figure — yours. Head terms in the US are easy. Long-tail queries in a small market, on mobile, with a specific postcode are where providers diverge. Test with your real keyword set, not with "pizza".
Parsing depth and honesty. Does the response include ads, People Also Ask, local packs, featured snippets, AI overviews, and their positions relative to organic results? And when the engine ships a new block type, how fast does the parser learn about it? A stale parser will silently drop a feature and shift every position below it, which looks exactly like a ranking change.
Location granularity. Country-level is standard and nearly useless for local SEO. City, postcode and named-location targeting is what you need for anything geographic. Check the actual supported list rather than the marketing claim.
Latency and the sync/async split. Live endpoints return in seconds and cost more. Queued endpoints return in minutes and cost substantially less. If you are running an overnight batch, the queue is free money — this is the single easiest saving available, and DataForSEO prices it at roughly a third.
Billing unit. "Per search", "per request", "per successful result", "per credit". These are not the same. Ask specifically whether failed requests are charged, whether one request means one page or one hundred results, and whether extra SERP features cost extra credits. A provider that charges only for successes and one that charges for attempts will differ by more than their prices suggest.
Rate limits and concurrency. A generous monthly allowance is worthless if you cannot spend it fast enough. If 45,000 monthly SERPs have to land inside a two-hour window, concurrency is the binding constraint, not volume.
The Test That Settles It
Every provider here offers a free tier or trial. Take fifty of your real keywords, in your real locations, and run them through two or three candidates on the same morning. Then compare the results to what you see in an incognito browser from the relevant location.
Count three things: how many requests succeeded, how many returned positions matching your manual check, and how many parsed the SERP features correctly. That comparison takes an afternoon and it beats every published benchmark, including the table in the next section — because it measures your queries rather than someone else's.
The APIs Compared
Prices are from each provider's own pricing page at the time of writing and are rounded for readability. They change frequently. Verify current rates at the linked pages before committing to anything.
| Provider | Model | Entry point | Effective per 1,000 | Free to try |
|---|
| DataForSEO | Pay as you go | Queued mode | $0.60 queued / $2.00 live | $1 signup credit |
| Oxylabs | Subscription | $49/mo Micro | from $0.50 | 2,000 results |
| Bright Data | PAYG or subscription | $1.50/1K PAYG | $1.50 PAYG / $1.31 on Scale | 5,000 records/mo |
| ValueSERP | Credit packs | $50/mo for 25K | $1.00–$2.00 by volume | Free trial |
| SerpApi | Subscription | $25/mo for 1K | $25 → $5.50 by tier | 250 searches/mo |
| Google Custom Search | Pay as you go | $5/1K | $5.00, capped 10K/day | 100 queries/day |
| Serper | Credit packs | Credit-based | Check current page | 2,500 queries |
Reading the Table
The effective per-1,000 column is the one that matters, and it spans an order of magnitude — roughly $0.50 to $25 depending on provider and tier.
DataForSEO is the cheapest credible option for batch work, at $0.60 per 1,000 SERPs in the queued mode and $2.00 in live mode. That queued-versus-live gap is the most consequential single number in this table: if your job can tolerate minutes instead of seconds, you pay roughly a third. Coverage is broad and the docs are detailed. In exchange, the API surface is large and takes longer to learn than the simpler competitors, and there is a minimum top-up on the account.
Oxylabs publishes from $0.50 per 1,000 results on its $49 Micro plan, dropping toward $0.40 at the $249 Advanced tier, with 2,000 results free to trial. Note their own caveat that effective pricing varies by target and whether JavaScript rendering is needed — the floor rate is not universal.
Bright Data sits in the middle at $1.50 per 1,000 on pay-as-you-go, falling to about $1.31 on the $499 Scale plan, and is explicit about charging only for successful requests. The free tier is the most generous here at 5,000 records a month, which is enough to run a small tracking job indefinitely without paying anything.
SerpApi is the most expensive per unit by a wide margin — $25 per 1,000 at the $25 entry tier, improving to $7.25 at $725/month and $5.50 at the largest published tier. What you get for the premium is the smoothest developer experience in the category: clean documentation, official client libraries, consistent response schemas across engines, and a reputation for keeping parsers current. If your engineering time is expensive and your volume is modest, that trade can be entirely rational. If your volume is large, it is very hard to justify.
Google's own Custom Search JSON API deserves a specific warning. It is official, it is documented, and at $5 per 1,000 queries with 100 free per day and a hard 10,000/day ceiling it is neither cheap nor scalable. More importantly, it queries a programmable search engine you configure, which is not the same index or the same ranking as google.com. It is not a rank tracking tool, and the number of projects that discover this after building on it is not small.
Where We Fit, Stated Plainly
Geonode does not sell a rank tracking API. We sell the proxies underneath one — residential from $0.79/GB with 1 TB free for new accounts, dropping to $0.50/GB at 100 GB and $0.27/GB at 1 TB, and datacenter from $0.14/GB, billed by traffic rather than per IP (current pricing).
For the straightforward job — "these keywords, these locations, parsed, daily" — one of the APIs above will beat what you can assemble from raw proxies, and we would rather say so than sell you a worse fit. The next section on build-versus-buy is where that changes, and it changes for reasons that have little to do with price.
Cost Modelling: What You Will Actually Pay
Headline prices are close to meaningless until you multiply them by your own workload. Here is a realistic one.
The scenario: 500 keywords, tracked across 3 locations, checked once daily. That is 1,500 SERPs a day, or roughly 45,000 SERPs a month.
| Provider | How it prices out | Monthly |
|---|
| DataForSEO (queued) | 45,000 × $0.0006 | ~$27 |
| Oxylabs Micro | $49 plan covers up to 98,000 | $49 |
| Bright Data PAYG | 45,000 × $0.0015 | ~$68 |
| ValueSERP | Two 25K packs, or the next tier up | ~$100 |
| Google Custom Search | 45,000 × $0.005 | ~$225 |
| SerpApi | 45K exceeds the 30K tier → Searcher plan | $725 |
Same data, same cadence: $27 against $725. A 27x spread that is completely invisible on the providers' front pages, because they are not quoting in comparable units.
Now Change One Assumption
Drop from daily to weekly checks. 500 keywords × 3 locations × 4 weeks = 6,000 SERPs a month, and the picture inverts:
- Bright Data: 5,000 free per month covers most of it; the remainder is about $1.50. Effectively free.
- DataForSEO: about $3.60 of usage — but there is an account minimum, so you are pre-funding rather than paying $3.60.
- SerpApi: 6,000 exceeds the 5,000-search Developer plan, so you land on Production at $150/month.
- Oxylabs: the $49 Micro plan, mostly unused.
At this volume the free tiers dominate and subscription minimums are what hurt. At 45,000 the per-unit rate is everything. The right provider genuinely depends on your volume, and the crossover is not where intuition puts it.
Three Levers That Change the Bill More Than Provider Choice
Cadence. Daily tracking costs seven times weekly tracking. Ask honestly how many of your keywords change position meaningfully day to day. Most do not. A common and effective pattern is daily checks on the fifty keywords that matter commercially and weekly on the rest — which often cuts the bill by 70% with no loss of information anyone acts on.
Queued rather than live. Where a provider offers both, batch mode is roughly a third of the price at DataForSEO. Overnight jobs have no reason to pay the live premium.
Location count. Locations multiply linearly and quietly. Three locations is three times the bill. Confirm each one is a market you actually make decisions about, rather than a row someone once added to a dashboard.
Apply all three and a $725 quote frequently becomes a $30 one before you have compared a single provider.
Build vs Buy
Having argued that most people should buy, here is the honest case for building — because it is real, and it is where proxies become the right purchase.
What Building Actually Involves
You need proxies with genuine geographic targeting, a request layer that manages sessions and pacing, a browser-fingerprint story convincing enough to survive detection, a parser for the results page, monitoring that catches silent failures, and a person who repairs the parser when the SERP layout changes — which it does, without notice.
That last item is the one people underestimate. The scraping is a weekend. The maintenance is forever. A managed API amortises exactly that cost across all its customers, which is why it is usually cheaper than doing it yourself even before you count your own time.
When Building Is Genuinely Right
Volume past the point where per-unit pricing dominates. Millions of SERPs a month, and the arithmetic changes. At that scale you are effectively paying a margin on infrastructure you could run.
Requirements no API models. You need a specific SERP feature nobody parses, a niche engine, a page layout as rendered rather than as parsed, or you need to follow through to the ranking pages themselves. Managed APIs give you their schema; if your question is not in it, you cannot ask it.
Location precision beyond what the APIs offer. Postcode-level or hyper-local targeting in markets the providers cover thinly. This is a real gap, and it is the strongest argument for residential proxies in this workflow.
Data residency or compliance constraints that rule out a third party seeing your keyword set. Sometimes your keyword list is the strategy, and handing it to a vendor is the thing you cannot do.
You are already scraping at scale. If the proxy layer, the parsers and the monitoring exist for other work, the marginal cost of adding SERPs is small and the build case is much stronger.
The Hybrid Most Teams Land On
Buy the API for the routine daily tracking, where reliability matters and the volume is predictable. Build for the specific, awkward, high-value queries the API cannot answer.
That is where proxies from a provider like us earn their place — not as a cheaper substitute for a SERP API, but as the layer under the 10% of work no SERP API covers. If someone tells you raw proxies are a straight replacement for a managed rank tracking API, they are selling proxies. We sell proxies, and it still is not true.
When You Do Not Need a Rank Tracking API
A section against our own interest and against most of this article. Several perfectly common situations need no API at all.
You want your own site's rankings, and only your own. Use the Google Search Console API. It is free, it is unlimited for practical purposes, and it reports actual average position from real impressions rather than a synthetic check from a synthetic location. It is more accurate than any scraper for this specific question, because it is measuring what really happened to real users rather than simulating one. The limitations are that it only covers properties you verify, only shows queries you already appear for, and lags by a couple of days. For a huge number of teams, that is the whole requirement — and it costs nothing.
You track fewer than about fifty keywords. A rank tracker with a dashboard is easier, cheaper and better at this than anything you would build. The API category starts making sense in the hundreds.
Nothing changes when the number changes. If rankings are reported monthly and no decision is ever taken on them, daily tracking is a cost with no corresponding action. Cadence should follow decisions.
You need competitive intelligence, not positions. Share of voice, content gaps and backlink data are different products. A SERP API returns results pages; it will not tell you why someone outranks you.
Your traffic is not from classic organic search. If acquisition runs through app stores, marketplaces, social or AI assistants, classic rank tracking measures a channel you do not depend on.
You are checking a handful of things occasionally. An incognito window with location settings is free and takes thirty seconds. Not everything needs to be automated.
The API becomes worth it at the intersection of scale, cadence and consequence: many keywords, checked often, where the number changes what someone does. Miss any of the three and you are buying infrastructure to generate a chart.
Common Mistakes
Comparing headline prices across billing models. $0.60 per 1,000 SERPs, $25 per month for 1,000 searches and $5 per 1,000 queries are three different units wearing the same clothes. Model your own volume first, then compare.
Paying for live when queued would do. Roughly a threefold difference for latency an overnight batch does not care about. This is the most commonly left-on-the-table saving in the category.
Tracking everything daily by default. Cadence is a multiplier on every other cost. Most keywords do not need it, and nobody acts on the difference.
Assuming a 200 response means correct data. The characteristic failure of scraped SERPs is a plausible page with subtly wrong contents. Validate a sample against manual checks periodically, not just when something breaks visibly.
Ignoring SERP features. "Position 4" above the fold and "position 4" below an AI overview, a local pack and four ads are not the same result. Tracking organic position alone will show a flat line through a collapse in actual clicks.
Using Google's Custom Search JSON API for rank tracking. It queries a programmable search engine you configure, not google.com. The numbers look reasonable and are not measuring what you think.
Building before checking Search Console. For your own properties, the free official source is more accurate than the paid synthetic one.
Forgetting the terms and the legal layer. Automated querying is generally against search engines' terms of service, and scraped data can carry personal data obligations depending on what you retain and where you are. Managed providers take on some of the operational risk; they do not take on your legal analysis. If this matters to your organisation, get advice rather than a blog post.
People Also Ask
What is a rank tracking API?
An HTTP service that takes a keyword, a location and a search engine, fetches the live results page and returns it as structured JSON — organic results with positions and URLs, plus SERP features like ads, snippets and local packs. It removes the need to handle proxies, bot detection and HTML parsing yourself.
How much does a rank tracking API cost?
Effective rates run from roughly $0.50 to $25 per 1,000 results depending on provider and tier. For a realistic workload of 500 keywords across 3 locations checked daily — about 45,000 SERPs a month — published pricing works out at roughly $27 with DataForSEO's queued mode and $725 with SerpApi. Verify current rates, as they change frequently.
Is there a free rank tracking API?
Several free tiers exist: Bright Data offers 5,000 records a month, Serper 2,500 queries, Oxylabs 2,000 results on trial, SerpApi 250 searches a month, and Google's Custom Search JSON API 100 queries a day. For your own site specifically, the Google Search Console API is free and more accurate than any of them.
Can I use Google's official API for rank tracking?
Not reliably. The Custom Search JSON API queries a programmable search engine you configure, which is not the same as google.com results, and it is capped at 10,000 queries a day at $5 per 1,000. It is a site-search product, not a rank tracking product.
Do I need proxies if I use a rank tracking API?
No. The provider handles address rotation, sessions and challenges — that is most of what you are paying for. Proxies become relevant only if you build your own tracker, or if you need location precision or SERP handling the managed APIs do not offer.
Why do rank tracking results differ from what I see in my browser?
Personalisation, search history, precise location, device, and the time of the check all shift results, and search engines run continuous experiments that show different users different pages. Some divergence is normal. Consistent, large divergence usually means your location parameter does not match where you are checking from.
Should I track rankings daily or weekly?
Weekly is enough for most keywords, and it costs a seventh of daily. Reserve daily for the small set where a position change would actually trigger a decision. Splitting the list this way commonly cuts the bill by around 70% without losing anything anyone acts on.
What is the difference between a SERP API and a rank tracker?
A SERP API fetches and parses live results and stores nothing — you own the scheduling, storage and history. A rank tracker stores keywords, checks them on a schedule, keeps history and reports changes, with an API for reading data out. Build on the first; buy the second if you mainly want a dashboard.
Wrapping Up
The useful decisions in this category are not really about which brand is best.
The first is which product you need. A raw SERP API means you own the scheduling, the storage and the history, and pay per request — right when you are building something. A rank tracker with an API means someone else owns all of that and you pay per keyword — right when you mainly need a dashboard. Choosing the wrong category costs far more than choosing the wrong vendor within it.
The second is cost modelling, and it is the step almost nobody takes. Multiply your own keyword count, location count and cadence by each provider's real per-unit rate before you look at a single feature list. Our worked example — 500 keywords, 3 locations, daily — ranges from about $27 to about $725 a month across published prices for identical data. Cadence, queued-versus-live mode and location count usually move the bill more than which provider you sign with, and all three are yours to control.
The third is whether you need to buy anything. If the question is "how does my own site rank", the Google Search Console API is free and measures real impressions rather than simulating a searcher. That is a better answer than anything on the paid list, and it is worth ruling out before spending.
We sell proxies, so the honest summary is this: for routine rank tracking, buy a managed API rather than assembling one from our product. Proxies earn their place underneath the awkward 10% — hyper-local precision, unusual engines, requirements no vendor schema models — and in the hybrid setup most teams eventually arrive at.
Whatever you shortlist, run fifty of your own keywords through two or three free tiers on the same morning and check them against a manual look. An afternoon of that beats every comparison table, this one included.
