Key takeaways
- In August 2026, sites with a Domain Rating of 46-75 earned about 46% of ChatGPT citations, according to Promptwatch Data. DR 61-75 was the single biggest band at 23.4%.
- DR 91-100 sites averaged only 4.4% for the month and sat near 3% from August 14 onward. That is less than DR 0-15 sites, which got 6.3%.
- Sites under DR 45 still earned almost a third of citations. Authority is not a gate.
- Third-party studies disagree with each other, mostly because of different methods, periods and scales (DR vs DA). Treat any single number as a snapshot.
- Practical move: benchmark against DR 46-75 competitors, build one page per distinct question, and aim PR at the DR 61-90 range rather than only chasing the biggest publications.
What the August 2026 data says
The cleanest first-party look at this question is Promptwatch's ChatGPT citation share by domain rank report, which covers ChatGPT citations from August 1 to 31, 2026. Domains with a known rank made up 89.65% of all citations in that window; the rest were excluded. Each daily bar adds up to 100% of the ranked citations.
The monthly averages by DR band:
| DR band | Average share of ChatGPT citations (Aug 2026) |
|---|---|
| DR 0-15 | 6.3% |
| DR 0-30 | about 14% |
| DR 0-45 | almost a third |
| DR 46-60 | 22.6% |
| DR 61-75 | 23.4% |
| DR 46-75 combined | about 46% |
| DR 61-90 | about 42% |
| DR 76-90 | 18.9% |
| DR 91-100 | 4.4% |
Two things jump out. The middle of the distribution is where the citations are. And the very top, the sites with the most authority on paper, got less than the bottom band.
The August 14 drop
The DR 91-100 story is not a slow fade. These sites averaged about 7% in the first week of August, dropped to roughly 5% on August 8, and landed near 3% on August 14, where they stayed. Most of the lost share went to the middle tiers: DR 61-75 gained 1.6 points and DR 46-60 gained 0.9 points.
August 14 is also the day reddit.com's share of ChatGPT citations collapsed from about 4% to 0.5%. Reddit is a very high-DR domain, so part of the DR 91-100 drop is probably just that one change. I would be careful about reading it as ChatGPT suddenly distrusting big sites. It is more likely that one huge source got dialed back and the share spread out.
There is also a second event worth knowing about. On August 8, ChatGPT Search started using the site: operator at scale, jumping from about 0.4% to about 17% of fanout queries. That is the same day DR 91-100 first stepped down. I can't prove the two are connected, but it is a plausible contributor, since site: queries pull results from specific domains rather than from a general authority-weighted pool.
What DR 46-75 winning actually means
A few readings are reasonable. Others are tempting but weak.
Reasonable: the middle is where the competition is
If almost half the citations go to DR 46-75, then the sites you are really competing with for a given prompt probably sit in that band. That makes it a better benchmark than comparing yourself to Wikipedia or a Fortune 500 domain. Pull your top ten competitors for your key prompts, check their DR, and see where you fall.
Reasonable: you do not need a top-tier DR to start
Promptwatch's own recommendation from the report is blunt: do not wait for a high DR before investing in AI search. A solid mid-authority site with relevant content is enough to get cited. DR 0-45 still earned almost a third of citations, so smaller sites are in the game too.
Weak: DR is irrelevant now
No. The data does not say authority is worthless. It says citations in August were spread across the middle bands rather than concentrated at the top. Other studies find authority still matters in different ways, which I get to below.
Weak: this will hold next month
It might not. This is one month, and the top-tier drop is tied to a specific event. Citation patterns in ChatGPT move. Promptwatch's citation drop report shows average citations per response falling around the GPT-5.3 rollout on March 4, 2026, and the half-life data shows how fast cited pages churn. Re-check the numbers every month instead of building a strategy on one chart.
What other studies say, and why they differ
The third-party picture is messier. That is useful, because it shows how much the answer depends on how you measure.
| Study | Scale | Headline finding |
|---|---|---|
| Promptwatch Data, Aug 2026 | DR | DR 46-75 earned about 46% of ChatGPT citations; DR 91-100 about 4.4% |
| Wellows, 20.5M citations across five engines, Mar-Jul 2026 | DA | 47.9% of citations came from domains under DA 40; 301 DA 90+ domains took 17.0%; ChatGPT shifted toward stronger domains in July |
| Green Flag Digital, finance pages, re-run Sept 2026 | DR | DR 90+ sites took half of citations across 999 finance domains, but through breadth of pages, not per-page strength |
| Ahrefs analysis cited by Remy | DR | 65.3% of cited pages with search visibility sat on DR 81+ domains |
| Position Digital, cited by eMarketed | Referring domains | Sites with 32K+ referring domains were 3.5x more likely to be cited |
These do not agree, and I don't think they have to. A few reasons:
- Different scales. Wellows uses DA and Promptwatch uses DR. The two are not interchangeable, so a DA 40 cutoff and a DR 45 cutoff are different slices of the web.
- Different engines. Wellows pools ChatGPT, Perplexity, Gemini and Google's AI surfaces. Google's AI surfaces were the most authority-heavy in that study, with a median cited DA of 47 to 50.
- Different topics. Green Flag looked only at personal finance, where banks, card issuers and government pages are naturally strong. Promptwatch's data spans all tracked prompts.
- Different periods. Wellows saw ChatGPT's median cited DA sit around 35 from March through June, then move into the low 50s by the end of July. Promptwatch's August window starts after that shift.
- Different questions. Ahrefs looked at pages that have search visibility. Remy's summary of the same research notes that 28.3% of top-cited pages had zero organic visibility, so a lot of cited content sits outside that sample.
When sources conflict, I weigh first-party, recent, ChatGPT-specific data most. That is the Promptwatch report. The rest are useful context, not contradictions to resolve.
Why authority buys breadth, not per-page strength
The Green Flag finance study has the most useful framing here. Across 999 finance domains, DR 90+ sites took half of all citations. But they did it because more of their pages got cited (median 524 versus 188 for sites under DR 50). Their citations per page were no higher.
That changes the strategy. You will not out-authority NerdWallet, but you do not need to. Every page that answers a distinct question is another ticket in the draw. Low-DR sites with hundreds of specific pages are already collecting citations, and the study found seven pages in the top 500 sitting on domains under DR 60, three of them under DR 10.
It also found that inside one already-trusted domain (1,000 Investopedia pages), referring domains showed no positive link to citations. Ranking breadth and topic demand did. So once a site is in the consideration set, building links to a single page does little. Picking a question many people ask and covering the many ways people phrase it matters more.
This matches the Wellows observation that the biggest July gainer in ChatGPT was a DA 38 specialist site. Deep topical coverage still counts.
Citations decay, so this is a maintenance job
Two more data points are worth holding onto. Promptwatch's content half-life report tracked every page ChatGPT cited on July 6, 2026 for 90 days. Landing pages and docs stayed cited longest, while news lost half its pages within 36 days.
Green Flag found something similar in finance: of the 215 most-cited pages in October 2025, 107 got zero ChatGPT citations eleven months later. A citation is rented, not owned. Winning once does not mean winning next quarter.
Promptwatch Data also shows what ChatGPT likes to cite in terms of format. Product pages were the most cited content type at 22.4% from Sep 28 to Oct 4, 2026, followed by landing pages and listicles. And almost half of ChatGPT responses cite 10 or more unique domains, which means there is a lot of room in each answer for a mid-authority source.
What to do if you are in the DR 46-75 range
- Benchmark the right competitors. List the domains cited for your target prompts and group them by DR band. If most are 46-75, you are on level ground.
- Build one page per distinct question. Breadth is how high-DR sites win. You can copy the pattern at smaller scale with focused pages on specific questions.
- Cover a topic from several angles. The Search Engine Land coverage of a domains study found that ChatGPT rewards domains that cover a topic from multiple angles, and that about 30 domains capture 67% of citations within a topic. Topical depth is how you get into that 30.
- Put the answer near the top. The same study found ChatGPT cited heavily from the upper part of a page.
- Aim PR at DR 61-90, not only the giants. Promptwatch's report suggests putting less weight on placements on the very biggest sites, since their share shrank in August. Brand-level trust still helps every page at once, so earned mentions on credible mid-tier publications are worth the effort.
- Refresh on a schedule. Given the decay numbers, plan to update pages that are earning citations before they fall off.
- Check that AI crawlers can read your pages. A page cannot be cited if the bot cannot fetch it.
Tools to track this
You can do a rough version of this by hand: run your prompts in ChatGPT and note which domains appear. It gets tedious fast, and DR bands vary by tool.
Promptwatch is the platform whose data this guide is built on. It tracks how ChatGPT and other AI engines cite your brand from the real interfaces, shows which of your pages get cited and where they land in the source list, and has a Content Agent for filling the gaps it finds. Its AI crawler logs help you tell an access problem from a content problem.
Other options in the catalog for monitoring AI visibility include Profound, Peec AI and Otterly.AI, which are mainly monitoring tools. If you want to pair citation tracking with traditional authority data, Ahrefs is the usual place to check DR.
Profound
Otterly.AI
For a wider list, see the GEO software directory.
| Need | Option |
|---|---|
| Citation tracking plus content creation and crawler logs | Promptwatch |
| Enterprise monitoring | Profound |
| Lightweight prompt tracking | Peec AI, Otterly.AI |
| Checking domain rating and backlinks | Ahrefs |
What I would not conclude
I would not tell a client that DR 91-100 sites have been dethroned. The drop in August lines up with Reddit's change and possibly the site: operator rollout, and Wellows saw the opposite direction a month earlier. I also would not say that DR does not matter. In finance, DR 90+ domains still took half the citations through sheer page count.
What holds up across every source is simpler. Authority helps a page get into the consideration set, relevance to the exact question decides a lot of what happens after, and no position is permanent. A mid-authority site with specific, well-structured pages and a habit of re-checking its data is in a good spot. If you want a quick read on where you stand today, run the free AI visibility report and compare your cited pages against competitors in your own DR band.