Brands seen buying newsletter ads grew 3.4 times from June to September, but ads per issue in a fixed set of 3,381 newsletters rose 9%

Findings

1. 2,015 brands had an ad in a newsletter issue dated 31 August to 29 September 2026, in 4,658 ads across 1,522 newsletters. In the last 7 days (23 to 29 September) the count was 712 brands, and in the last 90 days (2 July to 29 September) it was 3,977.
2. Brands seen per 30-day window grew 3.4 times between 2 June and 29 September, from 589 to 2,015. The number of newsletters read in those windows grew 6.8 times, from 7,522 to 51,137.
3. In a fixed set of 3,381 newsletters read in all four windows, ads per 100 issues went from 4.34 to 4.73, up 9.0%.
4. 91.9% of the 3,977 brands seen in the last 90 days were first seen in those 90 days, which reflects wider reading as much as new buyers.
5. In the last 30 days 83.7% of the 2,015 brands ran in one newsletter and 16.3% in two or more. The 95 brands in five or more newsletters (4.7%) account for 34.3% of the ads.
6. Among niches, AI brands were the most likely to run in two or more newsletters of the niche (26.5% of 219) and media brands the least likely (6.1% of 181).
7. Business was the only one of 11 niche groups whose ads per issue rose in every window in the fixed set, from 9.40 to 16.96 per 100 issues (180 newsletters). Health and finance fell in every window.
8. 26.6% of 6,760 brand and newsletter pairs seen in the last 90 days ran in two or more issues. 21.4% of the 2,292 brands first seen in July or August had an ad again in the last 30 days.
9. Reading the issues of the 10 busiest newsletters would show 6.6% of the 2,015 brands, and the 100 busiest 27.4%.

How to read these numbers. They count ads that Sponsor Radar found in newsletter issues it read. "First seen" means the earliest ad dated in that database, not the first time a brand bought an ad. Sponsor Radar read many more newsletters from August than in June and July, so every count of brands grows with the reading. Only the fixed set of newsletters and the shares within one window hold the reading constant.

What the last 7, 30 and 90 days show

About one brand in six ran in more than one newsletter, in every window. A small group does much of the buying: the 95 brands in five or more newsletters placed a third of the ads in the last 30 days. Note 03 found the same shape from June to September 22 (86.4% of brands in one newsletter).

Why counts of new sponsors mostly track wider reading

2026-06-02 to 2026-07-01589 brands

2026-07-02 to 2026-07-311,011 brands

2026-08-01 to 2026-08-302,045 brands

2026-08-31 to 2026-09-292,015 brands

Source: Carroway Research, Sponsor Radar database, issues dated 2 June to 29 September 2026. Newsletters read in the four windows: 7,522, 12,315, 31,066 and 51,137. Counts per window are in coverage-by-window.csv.

The count of brands rose from 589 in the 30 days to 1 July to 2,015 in the 30 days to 29 September. Over the same windows the database read 7,522 newsletters, then 12,315, 31,066 and 51,137. A brand can only be counted where an issue was read, so a rise in brands says little about whether more brands started buying.

To hold the reading constant, the table below follows the 3,381 newsletters that had at least one issue read in each of the four windows.

2026-06-02 to 2026-07-014.34 ads per 100 issues

2026-07-02 to 2026-07-314.87 ads per 100 issues

2026-08-01 to 2026-08-304.95 ads per 100 issues

2026-08-31 to 2026-09-294.73 ads per 100 issues

Source: Carroway Research. The fixed set is the 3,381 eligible newsletters with at least one issue read in each of the four windows. Ads and issues per window are in coverage-by-window.csv.

In the fixed set, ads per 100 issues rose 9.0% from the first window to the last. The rise came between the first two windows (4.34 to 4.87), and the rate was 4.95 in August and 4.73 in September. Fewer issues of these newsletters were read in the last window (7,098 against 15,706 in July), so its rate rests on fewer ads. The data supports a modest rise in ads per issue over the summer. It gives no basis for reading the 3.4 times rise in brands as a 3.4 times rise in buyers.

Which niches

ai26.5%

finance20.1%

sports15.3%

local13.4%

business12.7%

tech10.7%

culture7.8%

media6.1%

Source: Carroway Research. Niches with 100 or more brands in the last 30 days; the catch-all niche other is left out. A brand counts in each niche it appears in. Brand counts per niche are in multi-newsletter-share-by-niche.csv.

Brands that advertise in AI newsletters were the most likely to appear in several of them: 58 of 219 AI brands (26.5%) ran in two or more AI newsletters in the last 30 days, against 38 of 189 finance brands (20.1%), 29 of 229 business brands (12.7%) and 11 of 181 media brands (6.1%). Culture (7.8%) and media were the lowest of the niches with 100 or more brands. Education, productivity, climate, design and science, with 14 to 46 brands each, had no brand in two or more newsletters of the niche. Niche counts overlap, because a brand that ran in two niches counts in both.

Ads per 100 issues in the fixed set, by niche, for the 11 niche groups with 100 or more fixed-set newsletters:

Business is the one niche that rose in every window, from 58 ads in 617 issues to 78 ads in 460. Health (68 ads in 1,898 issues, then 5 in 871) and finance (55 in 582, then 20 in 445) fell in every window. The cells hold 3 to 109 ads each, so a move in one window is weak evidence and a move that repeats across windows is stronger.

Who comes back

Of the 2,292 brands first seen between 2 July and 30 August, 491 (21.4%) had an ad again between 31 August and 29 September: 157 of 768 July brands (20.4%) and 334 of 1,524 August brands (21.9%).

Within a newsletter, repeat buying is more common. Among newsletters with at least four issues read in the window, 1,795 of 6,760 brand and newsletter pairs (26.6%) ran in two or more issues over 90 days, and 583 of 2,671 (21.8%) over 30 days. The 90-day rate matches Note 03's 26.5% for June 1 to September 22.

How to find these sponsors

Sponsor Radar supplies the data in this note and appears in the table below as the example for the first route.

A newsletter has five ways to reach brands that buy ads. Each is scored on five criteria, with one point for Yes, half a point for Partly and none for No or not stated, and the routes are ordered by score. The criteria count equally, so a reader who values one more than another can re-rank the table. Routes with equal scores share a rank.

Each cell and its source page, read on 30 September 2026, are in routes-ranked.csv.

The busiest 1 newsletter1.1%

The busiest 5 newsletters4.3%

The busiest 10 newsletters6.6%

The busiest 25 newsletters10.5%

The busiest 50 newsletters18.8%

The busiest 100 newsletters27.4%

The busiest 250 newsletters41.7%

The busiest 500 newsletters59.0%

The busiest 1000 newsletters81.9%

Source: Carroway Research. Newsletters are ranked by the distinct brands each ran in the last 30 days, which a reader could only know after the fact. Brands seen per row are in reach-by-newsletters-read.csv.

Reading ads yourself is free and shows recent buyers, and exhibit 4 shows how far it reaches. Where each route falls short:

• Sponsor database with outreach. Book a Sponsor (called Sponsor Radar when this note was written) charges a $1 card check for 200 emails, then a plan from $19 a month, per its pricing page read on 24 October 2026, the newsletter still closes the deals, and its list holds the brands it has seen in the issues it reads. It does not take adult newsletters. Its public lists are the newsletter sponsors list and a weekly list of brands that bought ads.
• Reading ads yourself. The 10 busiest newsletters show 133 of the 2,015 brands, the 100 busiest 552, and the 500 busiest 1,188 (59.0%). That assumes you already know which newsletters carry the most sponsors. Reading gives names, with no contact and no price.
• Booking storefront or marketplace. Brands find you, so volume depends on the brands that look. Its creator page shows partner logos without dates, not recent buyers.
• Ad network inside your email platform. beehiiv's network is open to its Lite plan or above (its Ad Network page, read on 24 October 2026) and only runs on beehiiv. Kit's needs the Creator or Pro plan. Payouts are per click or per open, set by the network, and you choose among the offers shown.
• Sponsor database where you pitch. SponsorGap's Starter plan is $39 a month for 1,000+ sponsors with basic contact information. Its managed outreach starts at $999 a month plus a 10% success fee. Only its pricing page was read, and that page does not say whether the sponsors it lists bought recently.

Method

Corrected 24 October 2026. Two facts in the route comparison are out of date or were wrong: Book a Sponsor's price (it was 10 cents per email sent after $20 in free credits; it is now a $1 card check for 200 emails, then a plan from $19 a month) and the beehiiv plan its Ad Network needs (the Lite plan or above, not Scale). The counts, charts and scores are unchanged.

Source. The database behind Sponsor Radar, which reads newsletter issues and stores the sponsors it finds. The queries ran on 30 September 2026 against rows stored before 00:00 UTC on 1 October, with issue dates through 29 September, the last day with a full day of reading. 12,494 of the 12,885 ads are dated 2 June 2026 or later. The queries are in queries.sql. Ads are found by a language model and pattern rules; a hand check found 5 problem rows in 40.

What counts as an ad. A row linked to an issue that was read, dated by that issue's published date, that Sponsor Radar's public sponsors list would also accept: a tracking link, or a regex or sponsor-page source, or evidence text with a sponsor word or an opening "together with" or "partner message" header, and a paid confidence of at least 0.5 where one is stored. One brand in one issue counts once. Rows with no issue (sponsor-page logos and articles that name a sponsor) are left out, because their date is the day they were read. That removes 3,385 of 35,005 stored rows. 13,172 issue-linked rows pass the rules, which gives 12,885 ads from 4,428 brands in 2,725 newsletters (coverage-steps.csv).

Brands and newsletters. A brand is the registrable domain of the advertiser. Publishers, platforms, social networks, payment and government domains, investor hosts, country-code domains outside the US market and link-redirect hosts are excluded, as in Sponsor Radar's public list. Newsletters of the politics niche and on country-code domains outside the US market are excluded, in the ads and in the issues read. Sponsor Radar's adult-content screen and its merge of brands that share a name but not a domain are not applied, so a brand with several domains counts more than once. Sponsor Radar's public list counts more brands (6,008 on 30 September 2026) because it also counts sponsor-page logos and articles and undated rows.

Windows. Four 30-day windows ending 29 September: 2 June to 1 July, 2 to 31 July, 1 to 30 August and 31 August to 29 September. Last 7 days is 23 to 29 September, last 30 days is 31 August to 29 September, and last 90 days is 2 July to 29 September.

Fixed set. The 3,381 eligible newsletters with at least one issue read in each of the four windows. Ads per 100 issues is the ads in those newsletters divided by the issues of those newsletters dated in the window.

Niche. The niche stored for each newsletter by Sponsor Radar's classifier, with "other" as its catch-all. A brand's share in two or more newsletters is counted inside each niche.

First seen. The earliest issue date of a credible ad for the brand anywhere in the data. Return is a credible ad dated 31 August to 29 September for a brand whose first ad was dated 2 July to 30 August.

Pairs. A brand and a newsletter with at least one ad in the window, counted in newsletters with four or more issues read in the window, which is the rule Note 03 used. A pair repeats when the brand ran in two or more issues.

Routes. The five routes and the page for each were read on 30 September 2026. The score counts Yes as 1, Partly as 0.5 and No as 0. A feature a page does not mention is scored No, and a requirement it does not mention is scored as none. Open to any newsletter means no email-platform or paid-plan requirement.

Checks. In a reproducible sample of 40 ads from the last 30 days, read by hand against their stored evidence text, 35 were a sponsor message for the brand named. Two were not an ad (an editorial item about an ad product, and a call to partner with the newsletter itself), one could not be told from the text, and two were sponsor messages credited to a different domain than the advertiser's (an ad-server domain and a sister newsletter's domain). hand-check-40-ads.csv lists the 40.

Data

The CSV files below hold every number in this note. coverage-by-window.csv and active-brands-by-window.csv hold the windows and the fixed set. brands-by-niche.csv, multi-newsletter-share-by-niche.csv and niche-panel-by-window.csv hold the niche figures. repeat-pairs.csv and cohort-return.csv hold the repeat figures, and reach-by-newsletters-read.csv the reach of the busiest newsletters. placement-sources-last-30-days.csv and coverage-steps.csv trace the counts from the database, and hand-check-40-ads.csv holds the checked sample. routes-ranked.csv holds the route table with its sources. queries.sql holds every query. They are free to reuse with credit to Carroway Research.

• Coverage and ads by 30-day window (Exhibits 1 and 2) CSV, 4 rows, 0.5 KBOne row per 30-day window from 2 June to 29 September 2026: newsletters and issues read, newsletters with an ad, ads, brands and ads per 100 issues for all eligible newsletters, and the same counts for the fixed set of 3,381 newsletters read in every window.
• Brands with an ad in the last 7, 30 and 90 days CSV, 3 rows, 0.5 KBOne row per window: brands, ads, newsletters with an ad, brands in one and in two or more newsletters, brands in five or more and their share of ads, brands first seen in the window, and issues and newsletters read.
• Brands by niche in the last 30 days, and each niche in the fixed set CSV, 22 rows, 1.4 KBOne row per niche: brands, ads, newsletters with an ad and brands in two or more newsletters in the last 30 days, then the niche's fixed-set newsletters and ads per 100 issues for 2 June to 31 July and for 1 August to 29 September.
• Share of brands in two or more newsletters, by niche (Exhibit 3) CSV, 8 rows, 0.2 KBThe rows of brands-by-niche.csv for niches with 100 or more brands, other left out: brands, brands in two or more newsletters of the niche and the share.
• Fixed-set ads per 100 issues, by niche and window CSV, 44 rows, 2.3 KBOne row per niche and 30-day window for niches with 100 or more fixed-set newsletters: issues read, ads, brands and ads per 100 issues.
• Brand and newsletter pairs that ran in two or more issues CSV, 2 rows, 0.1 KBFor the last 30 and last 90 days, in newsletters with four or more issues read in the window: pairs, pairs in two or more issues, the share, newsletters and brands.
• Brands first seen in July or August that ran again CSV, 3 rows, 0.2 KBBrands first seen in each month and in both, how many had an ad between 31 August and 29 September, and the share.
• Brands shown by reading the busiest newsletters (Exhibit 4) CSV, 9 rows, 0.5 KBFor the 1 to 1,000 newsletters that ran the most distinct brands in the last 30 days: brands seen across them and the share of all brands.
• Where the last-30-day ads were read from CSV, 3 rows, 0.1 KBAds and brands by the extraction source stored with each row.
• Coverage and exclusions CSV, 10 rows, 0.7 KBEvery step from the stored rows to the ads, brands and newsletters analyzed.
• Hand check of 40 ads CSV, 40 rows, 5.1 KBA reproducible sample of 40 ads from the last 30 days with each row's stored evidence text and the verdict from reading it.
• Five ways to reach newsletter sponsors, scored CSV, 5 rows, 2.3 KBOne row per route: example, five criteria scored Yes, Partly or No, the score, and the page each cell was read from on 30 September 2026.