Skip to content
unnudged
Get early access
brandSteady

Similarweb on X

Similarweb is cited far more than it is discussed: its figures are the reference numbers in arguments about AI traffic share, news sites and app growth, with a smaller strand warning that what a figure counts has to be checked. The complaints are about whether the estimates can be trusted at all.

First written 27 Oct · what changed · from 100 posts by 83 people

Tracking posts that say “similarweb”

Track this for your market
Posts, last 24h
40
−47%vs. previous 24h
Posts, last 7 days
282
139 in the last 72h
Velocity
40/day
7-day average
Momentum
−1%
Last 24h vs. a normal day this week

Posts per day, last 30 days

as of Nov 3, 11:01 UTC
100500Oct 20: 42 postsOct 20Oct 21: 29 postsOct 22: 34 postsOct 23: 17 postsOct 24: 22 postsOct 25: 46 postsOct 26: 63 postsOct 26Oct 27: 53 postsOct 28: 32 postsOct 29: 47 postsOct 30: 23 postsOct 31: 30 postsNov 1: 81 postsNov 2: 26 postsNov 2

What changed

as of 27 Oct · checked once a month

First written 27 Oct. When what people say changes, it shows here.

Tone· 58 posts from 49 people· written 27 Oct

How do people feel about Similarweb?

Reception is mostly people using the numbers rather than judging the tool: 39 people cite a Similarweb figure as the evidence in an argument of their own. 8 use the data but warn about what a figure counts, and 6 recommend it.

The posts are mostly citations. 39 people treat a Similarweb figure as the number in an argument of their own: ChatGPT's share of AI web traffic slipping from 71.8% to 56.4% while Gemini reaches 25.0%, Grok Bot going from about 155,000 to 450,000 daily users in four weeks, Muse at about 2.18 million daily users, Truth Social down more than 30% in a year, half the 50 biggest English-language news sites losing more than a fifth of traffic in September. "Per similarweb" and "according to similarweb" are how the numbers get attributed.

8 people use the figures and flag the caveats: unique monthly visitors are not searches or sessions, two datasets can cover different windows and scopes, and a growth percentage means little without its base. 6 recommend it — one as the only reliable source for real user prompts, another as close enough when only relative movement matters.

  • Relies on its numbers41
  • Useful, with caveats9
  • Recommends it8
See 5 of 58 posts
J A Z I I@notjazii

this is best time for gemini 4 argon release according to similarweb, gemini lost over 1 percentage point of traffic share in the last 3 months even after google gave so many students free subscriptions, its share is still declining if team drops the gemini 4 family including argon, they will for sure cook harder and the rumor google gonna drop gemini 4 models tomorrow so let's see what happenes i just hope they don't delay it anymore

nick@thecsguy

grok bot nearly tripled its daily users in under a month, from about 155k on sep 20 to 450k on oct 17 per similarweb. i'm running around 19 bots and it still feels early. just a reminder, we aren't even at 1.0! what's the one task that made you open it every day? 🚀

David Hendrickson@TeksEdge

14.1% of Google users now also visit ChatGPT. But 93.7% of ChatGPT users STILL visit Google. At first glance it seems like ChatGPT isn't replacing Google but that's not what this Similarweb data seems to tell us. These are unique monthly visitors. So someone could use ChatGPT 500 times and Google once and still count as someone who uses both. Rather the charts seem to tell us → ChatGPT has penetrated 14.1% of Google's enormous audience → Google still reaches 93.7% of ChatGPT's audience What it doesn't really say is how many Google searches have those shared users stopped making? Because ChatGPT doesn't need people to abandon Google completely to hurt Google. It just needs to steal some of the queries.

Dennis Goedegebuure@TheNextCorner

🧭 Reading @kevin_indig and @aleyda back to back is the best kind of homework. Kevin's Growth Memo uses Similarweb estimates for 10,000 domains each in the US and France: paid search traffic up 21% and 17.3%, organic down 13.2% and 24.4%, August 2024 to August 2026. Aleyda's latest SERP Squeeze Monday, also Similarweb, shows paid clicks on "coffee tumbler" in Google US down from 3.9K to 2.3K, September 2025 to September 2026. My take: the window, the scope and the metric all differ. Two years vs one. Thousands of domains vs one query. Site-level paid traffic vs clicks on a single SERP. That's exactly why measurement teams should never borrow a headline number as their own benchmark. Open question: is paid still growing in aggregate while specific commerce SERPs start to squeeze it too? Grateful to have friends who publish the data that makes these questions possible.

Ethan Smith@ethan_graphite

How To Gather Real AI Prompts * There is much focus on how to optimize brands for particular prompts in AI. However, most people guess what prompts they should optimize for. This can call the entire strategy into question. * Outside of Similarweb, in our opinion, there are no reliable data sources that show users' real prompts. Existing tools are guessing based on dubious underlying data. * We propose a way to gather real user prompts in AI at scale for free. * Guide: link How Did You Hear About Us? * Attribution in AI is also hard. One solution is to ask users to self-report the channel primarily responsible for their conversion. This is often done by asking “How Did You Hear About Us?” * Self-reporting can also be extended to gather more information, namely, real user prompts. What Prompt Did You Use? We built a new conversion funnel at Graphite that asks the following questions. 1. How Did You Hear About Us? (channel) 2. What LLM Did You Use (product) 3. What prompt did you enter (prompt) This now tells you not only whether a user entered via AI, but also which LLM they used and the prompt they entered. Results * We now see 30 % of inbound leads come from AI. 52% came from ChatGPT and 40% from Claude. * We receive 10x more referral traffic from ChatGPT than Claude, while the % of leads that self-report Claude is almost as high as ChatGPT. This suggests conversions via Claude are particularly underreported using last-touch, far more than ChatGPT. * AI leads tend to be less qualified than other channels like referrals. * Only 21% of leads that self-report as coming from AI appear as referral traffic from AI. The rest came from search (49%) and direct (30%).

5 of the 58 posts behind this answer, one per person. The other 53, and every new one as it’s posted, come with tracking. Get early access

Pain points· 17 posts from 14 people· written 27 Oct

What do people complain about with Similarweb?

The complaints are about the figures: 11 people say the traffic numbers cannot be trusted. Billing and price are individual voices — one customer chasing a refund after a free trial, and two people who say the data needs paying for.

Doubt lands on the numbers themselves: traffic reported far above what free stat sites show — one post computes a 148-fold gap for a site cited in the press — bots inflating visits, samples called "cherrypicked", a crawler that site owners block, and estimates that miss a lot of small-site traffic. One post argues the browser-tab metric misses API calls, desktop clients and enterprise cloud usage, so a comparison built on it undercounts where the work happens.

The service complaints are individual. A student charged after a free trial posted three times the same day asking for a refund by 5pm. One person says you have to pay to see the data, another that a rival is free.

  • Numbers can't be trusted12
  • Billed after the free trial3
  • Too expensive2
See 5 of 17 posts
GoFakeYourselfMeghan@GoFakeYourselfM

Since I suck at math, I asked AI to average the difference between SimilarWeb’s traffic stats and the free website stats. (I chose SimilarWeb because that is who Newsweek sourced for their 144,000 figure.) Judging by AI's response, SimilarWeb's paid service reports traffic stats that are roughly 5x-10x higher than the free sites report. *cough* HOWEVER, according to Newsweek, SimilarWeb reports 144,000 unique visits per month for AsEver. com. That figure is: **148 TIMES higher than what the free websites' report!*** I REPEAT: 148x higher! 144,000 ÷ 974 = 147.84 Nowhere near the 5x-10x ballpark. *** Someone seems to be padding numbers for Meghan's shady business. It is way, way, way outside of the average(s) figured by AI and AI doesn't suck at math, like me. FTR, the free website, in my original post, does actually include worldwide figures. I don't know if it's just a select few countries, but it covers the ones that would be relevant to MeMe's sales...US, Canada, UK, etc. *Again, I cannot vouch for the accuracy of the free websites NOR SimilarWeb. I just wanted context for the number Newsweek reported.

Prabahan Kalita@prabahankalita

Traffic metrics (like SimilarWeb tracking ⁠claude vs openai⁠) measure browser tabs. They completely miss where serious technical work actually happens: - Headless IDEs & Agents: Cursor, Windsurf, Claude Code, and autonomous coding loops call Anthropic’s API directly. - Enterprise Clouds: A massive volume of Claude runs via AWS Bedrock and Google Cloud Vertex AI, which doesn't count as web traffic. - Desktop Clients & MCP: Claude Desktop and custom Model Context Protocol servers don't show up on browser leaderboard charts.

Kobe@kobe_ecom

$0 - $100k with ecom - day 85 lowkey did shit all today. did some product research, found a product, sent to supplier to see if they can source, meal prepped, and went gym. multiple hours spent on social media. ive got to keep myself accountable. i deleted ig and tiktok around 2 months ago and 2 days ago i downloaded them again for some fckn reason. they gotta go again. the only objections i have with this prod is that they can get similar at retail stores and a competitor doing very low numbers on similarweb but pretty sure it under estimates a lot and is inaccurate at small rev and they have over 100 ads running for like 90+ days so i think if i just speak better and be more relatable to my customers + grand slam offer i think i can pull it off. plus im kinda passionate about this prod, more than my others at least. also gonna run this a lot longer and test it way more thoroughly.

5 of the 17 posts behind this answer, one per person. The other 12, and every new one as it’s posted, come with tracking. Get early access

Alternatives· 11 posts from 9 people· written 27 Oct

What are the alternatives to Similarweb?

Semrush and cheaper routes are level at 3 and 3 people: people run Semrush alongside Similarweb for keywords, or get Similarweb data without the subscription — free tier, extension, shared accounts, a third-party proxy. Single posts name Ahrefs, SiteData, Google Trends and StatsCrop.

Semrush is one of the two things people weigh it against, level with the cheaper routes: 3 people put the two side by side, dropping small sites' domains into Similarweb for monthly visits and traffic sources and then checking Semrush for the organic keywords behind them, or listing free Trends plus paid Semrush and Similarweb as the working stack.

3 people describe the cheaper road: the free version is enough for a beginner with the Chrome extension, a shared account on a marketplace costs a fraction, and a third-party proxy includes Similarweb access.

The rest are one person each: Ahrefs now ships its own version of the traffic analytics and "absolutely no one needed" it; SiteData finds recently launched sites growing fast where Similarweb shows how big a site is; Google Trends is where one starts and Similarweb comes after; StatsCrop puts Japan at 4% where Similarweb has 47%.

  • Semrush4
  • Free and cheaper routes3
  • Ahrefs1
  • SiteData1
  • Google Trends1
  • StatsCrop1
See 5 of 11 posts
Jianyu Wu@iamjianyuwu

找了一天新词，截止目前还没找到合适的。 不过也在预料之中。上站对我来说不是门槛，有词的话可能半天就能上线。我真正在意的是：什么词值得做。 词选错了，后面越努力，离结果越远。 这两天把新词比赛四十多篇复盘都读了一遍，结合我自己的体会，说说我现在是怎么想的。 什么算新词 最近才冒出来，搜的人越来越多，但网上还没几个“像样的结果。 光有热度不够，背后还得有需求：一个能玩的游戏、一个能用的工具、一个能测的测试。只是一条新闻的，不算。 去哪找 越靠近源头越好。 最早的信号在社媒上：TikTok、X、Reddit、小红书，甚至朋友圈。大家在讨论什么、在问「这个在哪玩」，那就是新词。再往后是 GitHub 新项目、应用榜、Steam 新游戏。然后才轮到 Trends 词根和 Similarweb 找站。 复盘里好几个第一名，词都是刷社媒刷到的。原因很简单：每个人刷到的东西不一样，撞车的少。而 Trends 上的词，你看得到，别人也看得到。 所以我现在的看法是：Trends 拿来验证，不拿来发现。 怎么判断该不该做 主要通过下面的五件事： - 还在涨吗？热度能不能接近 gpts？ - 为什么火？是主播、梗、版本更新，还是别的？ - 竞争大不大？intitle 结果几百以内，域名还没被抢光 - 现在的结果有没有把需求接住？别人只是 iframe 一下的话，就是机会 - 来不来得及？新词一般一周以内见分晓，晚了就放手 KD 这类指标对新词感觉没什么用，词太新，工具里还没数据。自己看不懂的词不做，有侵权风险的词也不做。 怎么算做得好 上线一两天还没流量，大概率是词不对，换下一个。 有流量之后，看用户留不留得住：停留时间、跳出率，还有没有人主动把你的站分享出去。 最后看能不能挣到钱。很多人复盘时最后悔的，都是付费上得太晚。 还有一条最容易被忽略：做完一个词，下一个词有没有变得更容易？新词本来就是概率游戏，十个站能起来两三个就算不错了。能靠的是方法越磨越顺，而不是哪一次运气好。 今天虽然词没找到，也算是摸了摸新词的门道 大家平时有更科学有效的方法吗？或者我上面的理解跟真实场景有没有gap？ 欢迎前辈们指点迷津呐🙏

Victor在西雅图@victor_UWer

逆天，判断一个 AI 模型值不值得做套壳站，我只看搜索结果页前三名。 贼简单，三步： 1. 隐身窗口搜 "kling 2.1 free online"，换成最近刚发布的模型名也行 2. 看前 10 条。官方站排第一，后面全是新闻和评测，就别碰了。要是前三里混着几个小工具站，说明官方没把这个词守住 3. 把这些小站的域名丢进 Similarweb，看每月访问量和流量来源。搜索占比高，说明流量是靠这个词吃来的，不是靠投放 再多做一步：去 Semrush 看它 Organic Keywords 里有没有别的模型词，有的话它是在批量铺词，你也可以照着铺。 新模型刚发布的头两周，结果页最乱，位置最好抢。

yikun@yikun900655

新词笔记 #005：Similarweb 找到一个新词，搜一下发现已经有人做了站，这个站到底有没有流量？Similarweb 就是偷看别人网站流量表的工具：月访问量、流量来源（搜索还是直接访问）、主要国家，一查就知道。同类站真有人来，这个词才值得做。 我的理解：免费版够新手用，装个 Chrome 插件，顺手一点就能看。 官网：link 你做站前会先看竞品流量吗？

5 of the 11 posts behind this answer, one per person. The other 6, and every new one as it’s posted, come with tracking. Get early access

Questions· 15 posts from 12 people· written 27 Oct

What do people ask about Similarweb?

The largest group of questions, from 5 people, is about what a figure counts before it can be used: deduped visitors or sessions, bots or humans, usage or installs. 3 ask for the number behind a trend, 3 how the estimates are made, and 2 whether web and app, Android and iOS, are separated.

The questions are short and put to the account. 5 people ask what the number is: deduped monthly visitors or raw sessions, what happens with bots, whether a chart shows usage or installs, whether 3.35% growth is a ceiling or traffic spreading to other tools. One notes the percentages rest on different denominators, so a gap alone does not show one service replacing another.

3 people want the next cut: day-7 retention for Gemini alongside the DAU rise, how many are still active a month after their first bot, whether people who use both services run fewer Google searches. 3 ask how the estimates are made — one sets a Similarweb weekly figure of about 440,000 against 800,000 confirmed elsewhere, another asks why a store with hundreds of active ads does not appear at all. 2 ask about platform splits: an Android-only cut, when it lands, and whether web version and local app are told apart.

  • What the figures actually mean5
  • Can you break that down further?4
  • How the estimates are made3
  • Web or app, Android or iOS?3
See 5 of 15 posts
warriorwaypoint@warriorwaypoint

What is this telling me? Usage or installs. I’d be curious to know what usage looks like. Also Muse is free and you can pay for more while Grok @bot you are paying for usage. What are the paying users of Muse, do you have that?

5 of the 15 posts behind this answer, one per person. The other 10, and every new one as it’s posted, come with tracking. Get early access

Wishes· 6 posts· written 27 Oct

What do people want from Similarweb?

4 people want the counts behind the percentages: absolute visits alongside the shares, and how many of the new users come back. The other asks are single voices — a relatively neutral traffic read, and a plan that fits the budget.

The section is small. The biggest ask, from 4 people, is for real numbers next to the shares: whether a chart shows absolute visits too, since a high new-user share looks the same whether a product is growing fast or people are not returning; how many are still active a month after their first bot; how many of the new users turn it into a daily tool; and age shares paired with user counts to show the scale of the audience.

One person asks how to get a relatively neutral traffic read when the estimates are biased. One, who calls the product amazing, says they need a better plan.

  • Pair shares with real numbers4
  • A more neutral read on traffic1
  • A plan that fits the budget1
See 5 of 6 posts
Holleazy@holleazyai

That Similarweb chart goes from ~155K to ~450K daily users between Sep 20 and Oct 17, close to 3x in four weeks. The number I'd want next is how many are still active a month after their first bot.

5 of the 6 posts behind this answer, one per person. The other 1, and every new one as it’s posted, come with tracking. Get early access

In their words

How do people talk about Similarweb?

The phrases people use most, counted once per person.

  1. 1“daily users”7
  2. 2“similarweb data”4
  3. 3“traffic share”3
  4. 4“per similarweb”3
  5. 5“similarweb estimates”2
  6. 6“daily active users”2
  7. 7“monthly visitors”2
  8. 8“google searches”2
  9. 9“similarweb chart”2

Posts are shown the way X shows them. One of them yours? Here’s why it’s shown, and how to take it off.

Also in SEO suites

Get every post about Similarweb, filtered and labelled

This page is the summary. Tracking gives you the stream behind it for any brand, keyword or @mention: each new pain point, request and question as it’s posted, with the spam and promotions already taken out.