DR3 vs DR4: what changed across 100 project domains
We compared DR3 and DR4 on 100 selected VCL project domains on November 9, 2026. 93 scored lower, five were unchanged and two scored higher. Here is what that does—and does not—tell founders.
On November 9, 2026, we compared DR3 and the DR4 preview for 100 selected custom root domains from public VCL project listings. 93 scored lower under DR4, five were unchanged and two scored higher. Domain Rating by Ahrefs, obtained through Detailed's DR4 preview.
This is a comparison of two models at one point in time. It does not show that those businesses lost visitors, customers or product quality.
What we found across 100 project domains
All 100 selected domains returned both readings. The sample median was 17 for DR3 and 0 for DR4; 65 domains had a DR4 reading of zero. Domain Rating by Ahrefs.
Domain Rating by Ahrefs. Measurement window: November 9, 2026, 08:17:59–08:20:53 UTC. This is a selected sample, not an estimate for all websites.
| Score band | Domains under DR3 | Domains under DR4 |
|---|---|---|
| Exactly 0 | 3 | 65 |
| Above 0, below 10 | 28 | 22 |
| 10 to below 30 | 38 | 6 |
| 30 to below 50 | 25 | 4 |
| 50 to 100 | 6 | 3 |
| Total | 100 | 100 |
Domain Rating by Ahrefs. Download the aggregate counts (CSV). The sampling and measurement methods are documented below.
The difference between the two sample medians is 17 points. That is not the median change for individual domains and should not be reported as such. The 93/5/2 counts compare each domain with itself, making them the clearer summary of direction within this sample.
How we selected and measured the sample
We retrieved public VCL project listings in newest-first order and selected the first 100 unique eligible custom root domains. Four listings were excluded by the shared-host eligibility check before reaching that target. We did not select domains because their DR readings were high, low or surprising.
Each website was normalized to its registrable root domain. That means this is a root-domain cohort, not a test of the update's treatment of hosted subdomains. Root-level measurements can include activity outside the listed app itself.
We checked the selected domains sequentially through Detailed's public preview, retaining timestamps and both model readings. There were 100 successful pairs and no failed checks. All calculations use those 100 pairs. The measurement window lasted less than three minutes, but the readings are still a dated snapshot rather than permanent scores.
We publish aggregate results and methods rather than a downloadable database of individual Ahrefs scores. The chart and table can be cited with attribution to VCL's study and to Ahrefs for Domain Rating.
What the sample cannot tell us
- It is not representative of the whole web. VCL listings attract particular kinds of builders, and newest-listed does not mean newly registered.
- It does not diagnose any individual link profile. We did not classify the links behind every score or label the sampled businesses as spam.
- It does not measure lost traffic. No traffic or signup change was inferred from a score difference.
- It does not prove which service increases DR4. We did not run a link-building intervention or a controlled comparison of providers.
- It does not establish a permanent conversion between versions. A DR3 reading cannot be converted into a DR4 reading with a fixed formula.
For the actual model changes and primary sources, see our DR4 explainer. For business decisions, keep the score beside your own search, referral and conversion data.
What founders can do with these findings
Record a fresh baseline with the host, model and observation date. If you compare competitors, compare like with like. For hosted projects, check the exact host rather than substituting a platform's root domain.
Then review your next acquisition action on its own merits. Would a directory listing reach the right audience? Could a partner integration or useful public resource earn a relevant mention? What would count as a qualified signup? Our Domain Rating improvement guide provides a practical sequence.
Keep old reports intact and label their model. Rewriting history to make a chart look continuous would hide the very change you need to understand.
What we are changing at VCL
Our checker labels DR3 and DR4 separately, and our existing historical results retain their DR3 context. New paid DR Growth campaigns remain paused while we assess the new baseline and what providers can substantiate. A free comparison is available without joining a waitlist.
Want help with your site's growth when offers reopen? Join the buyer waitlist. Joining does not fund a campaign or guarantee availability, rankings or traffic.
Can you provide credible SEO services? Join the provider waitlist. Prepare evidence with exact hosts, dates, metric versions and disclosed methods. A score difference between two models is not a delivery case study, and waitlist signup is not approval to sell.