Neither total answers the cancel-Ahrefs question on its own. DataForSEO's backlink API page describes a continuously crawled index of roughly 2 trillion backlinks counted as live, meaning still present, with history back to 2019. Ahrefs' Backlink Checker page claims the largest live index in its class at over 15 trillion live links, and the comparison pages put its full external total around 35 trillion, with about 500 million domains tracked and a refresh every 15 to 30 minutes. Whether the smaller index trails or beats the larger one on links that still exist is disputed in public, which is why audits that start from totals keep stalling.
A backlink, a link from one site to another, only helps while the linking page still exists and still links. A referring domain, one unique website linking to a target, only counts while at least one of its links is live. Totals mix live links with history. An audit needs the live share.
What 2 Trillion vs 35 Trillion Actually Counts
The two live counts do not use the same bucket. DataForSEO's index pages break its side into about 2 trillion live external backlinks; 192 to 336 billion live pages depending on how lost links are counted; about 289 million domains with at least one active backlink; daily crawls of around 8 billion links and 1 billion pages; a full recrawl cycle of up to 90 days with continuous additions; up to 4 trillion links when lost backlinks kept for a year are included; and over 6 trillion with the deeper archive. History runs back to the start of 2019 through dedicated endpoints.
Ahrefs' pages put its side at roughly 35 trillion external backlinks and about 500 million domains, with the live index rechecked every 15 to 30 minutes, and the checker page states the over-15-trillion live figure as its live claim.
Four endpoints shape what you can pull from the DataForSEO side: Backlinks (a target's backlinks and related data), History (back to 2019), Anchors (the clickable words of each link), and Summary (an overview of a target's profile). Its scoring mirrors Google's original PageRank, the early method of weighting a page by the links pointing at it, with a formula DataForSEO documents for backlinks, domains and pages. Integrations listed on the same page include Make, Google Sheets and n8n.
DataForSEO's own comparison post warns against trusting aggressive marketing or misleading backlink index studies, and does not publish the controlled test that would replace them.
Live Links vs Totals: Live Ratio, Lost Handling and Refresh
Live ratio is the share of returned links that still resolve: the page returns 200 rather than 404, and the link is still present in the HTML with its attributes intact, including nofollow (a tag telling search engines not to pass credit) and UGC (a marker for user-generated content such as forum posts). Lost handling is how a vendor keeps or drops links that disappeared, and for how long. Refresh is how fast a re-crawl picks up new and lost links.
That is where 2 trillion against 35 trillion misleads. A larger total can hold more history. A smaller live index can hold a higher live share on a given sample. No published fixed-sample test, with a live definition, endpoint settings and a date, lets an operator compare live ratio directly, which leaves you to verify live status yourself.
Why the Same Database Gets Called Trash and Outperforming
From late August into mid-September the same ceiling question keeps returning: can cheaper pay-per-use data replace a $129-a-month suite for backlink depth. On one side the words are trash, poor, way behind, bad data, with backlink depth named as the reason a cancellation stays on hold. An early cancellation thread carries a sharper report, high-authority-by-DataForSEO links that did not survive a manual audit, from one account with no audit sheet attached.
The counter-claim arrives secondhand. Mid-September replies cite a 2026 Habr experiment across hundreds to as many as 10,000 domains in which DataForSEO and Semrush often returned more unique live referring domains, with a higher live share, than the Ahrefs API in some tests. Cited, not reproduced.
Two related notes recur around those threads. Builders ask whether most tools beyond Ahrefs and Semrush are wrappers around one data provider, with those two described as running their own large crawlers and independent indexes. And one September account runs the opposite way: DataForSEO numbers feel inflated next to Ahrefs' more conservative read, with keyword examples where one side showed tens of thousands and the other hundreds. Totals without a live definition cannot tell you which number to build on, which is the only thing all of those accounts agree on.
What an Audit Costs at Per-Row Math
DataForSEO's backlink pricing bills live pulls at $0.024 per request plus $0.000036 per returned row, where a row is one object describing a single backlink, referring domain or page. A request that returns 1,000 rows therefore costs $0.060, with results in up to two seconds in live mode. Ahrefs anchors the other side at Lite from $129 a month, with metered API units that scale with depth and endpoint complexity.
Use that to price before you pull.
| Audit size | DataForSEO at $0.06 per 1,000-row request | What to compare it against |
|---|---|---|
| 10,000 backlink rows for one competitor | about $0.60 | one slice of a $129 month before units |
| 100,000 rows across ten competitors | about $6.00 | whether that month would have covered the same ten pulls |
| 1,000,000 rows for a full referring-domain sweep | about $60.00 | whether repeated sweeps make the subscription cheaper than pay-per-pull |
Row math is not the full audit. Deduplicate domains before you expand rows, keep new and lost separate through the history endpoints, and record endpoint, parameters and date on every pull so a later recheck compares like with like.
Open Crawl Plus Vendor Enrich: The Hybrid Live Check
A mid-September French-language thread describes the lightest pattern that answers the live question without trusting either total. The builder starts from Common Crawl, an open crawl of the web that anyone can process, as the primary graph. Exact source URLs come from analyzing that full base on demand. DataForSEO is then added for fuller URL coverage, with spam removed. Authority math stays on the Common Crawl side, where the inputs are visible.
A test of that setup on one site, paulvengeons.fr, reported the fields an audit actually needs: the exact pages where each backlink was found, destination URLs, links still present, links removed, misconfigured links pointing at a 404, nofollow and UGC markers, and a status per backlink. The write-up highlights one misbuilt link that looked fine as a referring-domain count but pointed at a missing page and lost its value. One builder's setup, with no independent check of how much of the web it recalls, and still the clearest description of what to verify.
Treat any vendor database as enrichable. Pull the total, then verify existence, destination and attributes before counting a link toward a decision.
How to Audit without Trusting Either Total
Fix a small sample first. Pick five to twenty domains you care about, including your own and two competitors whose backlinks you can already partly see.
Pull both sides with the same rules. Record target, endpoint, parameters and date. Keep live and historical separate, live for what resolves now and history for what was lost and when. Keep referring domains separate from total backlinks, since one domain can carry many links.
Verify live status before scoring. Check each linking URL for 200 against 404, whether the link is still present, where it points, and whether it carries nofollow or UGC. For a handful of URLs, a dated live check of Google's results shows whether the linking page is still indexed at all, which is the part qequals covers with its weekly snapshots; the link-graph depth itself stays with the backlink vendors. Apply Domain Rating-style scores, third-party authority numbers in the style of Ahrefs' metric, only after the live check.
Price the full run from the sample. Multiply verified rows by the per-request rate to estimate the DataForSEO side, and compare against the subscription months and unit burn the same pulls would consume on the suite side. If the verified-live list does not change which links you would keep, fix or discard, pause the pull. The audit is finished when the live list stops moving your decisions, not when the totals agree.
