# Backlinks: overview, dated links and history (for backlink-vs-traffic studies)

Three FetchAPI endpoints answer "how many backlinks does this site have, which ones, when did they arrive, and how did the
count grow". They were chosen in a 2026-11-13 benchmark of 9 sources on 12 domains (big, mid and small). Accuracy was
tested by fetching 75 listed links per provider and checking each page still links. Raw data and scripts:
`.context/backlinks-bench/`.

Header: `x-api-key: <key>` (an owner key for the scraped Moz/Ahrefs fallbacks). Each answer's `meta` shows the provider
and `cost_usd`.

## 1. Overview: totals for many domains

```
GET https://fetchapi.co/api/v1/web/backlinks?domains=linear.app,resellbot.com
```

Up to 1,000 domains a call; ~$0.0012 a domain + $0.035 a call; ~10 s.

Per domain:
- `referring_domains`, `backlinks`;
- `spam_score` (0-100);
- `authority_score`.

linear.app (2026-11-13): 11,280 referring domains, 151,543 backlinks, spam 11.

## 2. Top backlinks with dates

```
GET https://fetchapi.co/api/v1/web/backlinks?domain=resellbot.com&top=100
GET https://fetchapi.co/api/v1/web/backlinks?domain=resellbot.com&top=500&list=referring_domains
```

`list` options:

| `list` | One row per | Fields |
|---|---|---|
| `backlinks` (default) | linking domain (its best link) | `url`, `domain`, `anchor`, `dofollow`, `first_seen`, `last_seen`, `domain_rank` (0-1000), `spam_score` |
| `all_links` | link | same as above |
| `referring_domains` | linking domain | `domain`, **`first_seen` (when that domain first linked)**, `lost_date`, `backlinks_from_domain`, `referring_pages`, `domain_rank`, `spam_score` |

The summary also includes `referring_ips`, `referring_subnets`, `dofollow_referring_domains_pct` and `links_first_seen`.

Price: $0.03 a call + $0.05 a domain + $0.0008 a row. resellbot.com, top 10: 13 s.

## 3. History: monthly counts back to 2019

```
GET https://fetchapi.co/api/v1/web/backlinks-history?domains=linear.app,resellbot.com&months=24
GET https://fetchapi.co/api/v1/web/backlinks-history?domains=linear.app&from=2019-01-01
```

Up to 100 domains a call; $0.035 a domain + $0.0015 a month: **$0.071 a domain for 24 months**. 12 domains x 22 months
took 9 s.

Per domain, `series[]` holds one entry a month (`month` = calendar month, measured on its last day):
- `referring_domains`, `referring_main_domains`;
- `backlinks`, `referring_pages`;
- `referring_ips`, `referring_subnets`;
- `nofollow_referring_domains`, `authority_score`;
- `referring_domains_change` (vs the previous month).

## Quality: what the benchmark found

| Source | Live-link precision | Recall of verified linking domains | Per-link dates | History | $ per domain |
|---|---|---|---|---|---|
| **DataForSEO-backed (cheapapi, steady-scrape)**: used for all three endpoints | **86%** | 62% | first, last, lost (first seen a median 35 days after the page's publish date) | monthly from 2019, smooth | $0.0012 counts; $0.07 for 24 months |
| Moz (radeance): fallback for lists | 57% | **83%** | last seen only | 30 days | $0.006-0.009 |
| Ahrefs free checker (s_actors): last fallback | 68% | top 20 only | none | big sites only | $0.009-0.011 |
| Semrush (pro100chok) | 43% | partial | first, last, lost (since 2025-08) | 6 months | $0.0045 |
| Common Crawl web graph (free) | n/a (domain level) | 34% | none | 55 releases, noisy (direction check only) | free |
| SpyFu | 0 backlinks on small sites | | | | |

**Referring domains, three sizes of site:**

| Source | linear.app | resellbot.com | sold-comps.com |
|---|---|---|---|
| DataForSEO-backed (ours) | 11,279 | 208 | 76 |
| Moz | 15,601 | 214 | 261 |
| Ahrefs | 17,539 | 1,635 | not measured |
| Semrush | 24,912 | 2,573 | 1,964 |

Semrush and Ahrefs report about 10x more on small sites. On flowlister.com, 1,650 of their linking domains shared one
link-building spam anchor, so do not read their small-site counts as real links.

**History agrees with other sources in direction:**
- calendly.com over a year: ours −5.4%, Ahrefs −5.6%.
- tally.so over a year: ours +43%, Ahrefs +75%.
- Ours and Common Crawl moved the same way on 6 of 7 domains over two years.
- Month-to-month moves are about 4% on mid and big sites.

## Recipe: do backlinks (and backlink growth) drive traffic?

1. **Traffic per month:**
   - `POST /api/v1/web/traffic-bulk` (`ai=1` adds organic-search share) gives 3 months.
   - `GET /api/v1/web/traffic-history?domain=` gives 5-12 years (Similarweb, via Dealroom).
   - Both are calendar months, like the backlink history.
2. **Backlinks per month:** `web/backlinks-history` with `months=24`, 100 domains a call.
   - **Cost:** 1,000 domains is ~$71.
   - **Add ~50 stable control domains** (large, long-established sites) to remove index drift.
3. **Arrival dates** (optional, for event studies): `web/backlinks?list=referring_domains&top=1000` gives when each
   linking domain arrived (`first_seen`).
   - **Lag:** first seen trails the real date by a median of ~1 month.
   - **Cost:** ~$0.30 a domain at 300 linking domains.
4. **Analysis:**
   - Correlate the monthly change in referring domains with the change in visits, lagged 0, 1, 2 and 3 months.
   - Use organic-search visits (visits x `search_organic` share) as well as total visits: links should move search
     traffic first.
   - Weight by `dofollow` and the linking domain's `domain_rank` if you build a quality-weighted link count.
5. **Snapshot for every domain:** `web/backlinks?domains=` (1,000 a call, ~$1.24 for 1,000).

## Known weaknesses (put these in the study's limitations)

- **Index jump in late 2026:** every domain jumps **+14% to +360% between 2026-09 and 2026-11** in the DataForSEO-backed
  history. That is an index change, not growth. **End the series at 2026-09**, or divide by the control panel's median
  growth each month.
- **Missed links:** the DataForSEO-backed index misses about **38% of real linking domains** (recall 62%). Absolute
  counts are low; changes over time are the reliable part.
- **Dates lag:** `first_seen` is when the crawler found the link, a median of 35 days after the page was published (7 of
  44 sampled were earlier than publication).
- **Different indexes, different counts:** never mix them in one series. Ahrefs and Semrush inflate small sites; Moz keeps
  more dead links (43% not live).
- **Licence:** these are scraped indexes (DataForSEO, Moz, Ahrefs). Use them internally; don't republish the raw numbers.