To drive traffic from Twitter/X in 2026, deliver the value in the post itself, use images or text instead of a bare link, put the URL in a reply or your profile (or post from a Premium account), tag every link with UTM parameters, post on weekday mornings and afternoons, and publish consistently so followers learn to click.
X used to be one of the best referral sources on the web. A sharp headline and a link could send thousands of visitors to a blog post in an afternoon. That era is over for most accounts. Buffer's analysis of 18.8 million posts found that since March 2025, the median engagement rate on link posts from regular, non-Premium accounts has been "stuck at zero". The open-source code for X's feed, updated in 2026, shows a ranking system that predicts more than a dozen actions per post and discounts content from accounts a viewer does not follow. Traffic from X is still achievable, but only with a playbook built for how the feed works now. This guide covers what changed, how traffic differs by account type, format, link placement, content goal and industry, the timing data, how to measure, a step-by-step method and the mistakes that waste the most clicks.
Key findings
- Link posts are the weakest format on X. Buffer's 2026 content format analysis put the median engagement rate for link posts at 2.25%, against 3.56% for text, 3.40% for images and 2.96% for video.
- Regular accounts' link posts are effectively invisible. In Buffer's dataset, regular accounts' link posts sat at a 0% median engagement rate as of August 2025, while Premium accounts' link posts reached about 0.28%.
- Premium buys reach. Premium accounts saw roughly 10x the median reach of regular accounts: under 100 impressions per post for regular accounts, close to 600 for Premium and over 1,550 for Premium+.
- Reach on X is shrinking overall. Platform-wide median reach fell from near 1,000 in August 2024 to under 750 in August 2025.
- The algorithm predicts link clicks. X's open-source feed code lists "link" among the click actions its ranking model predicts, alongside replies, reposts, dwell time and negative signals such as "not interested".
- Every URL costs 23 characters. X counts any valid URL as 23 characters, whatever its length, so long tracking URLs do not eat into your 280.
- Weekday business hours lead. Sprout Social's 2026 data points to Tuesday to Thursday, 12 to 6 p.m. local time; Buffer's analysis of 8.7 million posts found 9 a.m. on Tuesday to be the single best slot.
What's changed in 2026
The feed code is public again, and it is a prediction engine
X's For You feed code now lives in a public repository. Its README describes a feed "assembled per request" from in-network and out-of-network posts, ranked by a transformer model called Phoenix that predicts how likely each viewer is to take each action. The final score is a weighted sum of those predictions: favorites, replies, reposts, quotes, shares, clicks (including link clicks), dwell time, follows, minus negative actions such as "not interested", mute, block and report. The repository's major release is dated 13 August 2026. Two details matter for traffic: posts from accounts a viewer does not follow are multiplied by a discount factor below 1, and later posts from the same author are discounted for diversity. In plain terms, you earn distribution by making people stop, read and act, and you reach non-followers only when the post is strong enough to overcome the discount.
Link suppression became measurable
X has not published a link penalty, and the code does not show a simple "links minus X" rule. But the outcome is visible in the data. Buffer reports that links were already the lowest-performing format in March 2025, and after that point fell to absolute zero for regular accounts, while every other format kept some engagement. Premium accounts still get reach on link posts, but links underperform text, images and video even for them.
Premium became a distribution tier
Buffer's study of 71,000 accounts found about 27% subscribed to a Premium tier, and those accounts saw roughly 10x the median reach of regular accounts. For anyone whose X strategy depends on traffic, Premium is now a cost of doing business, not a vanity badge.
X now has the lowest engagement of the major networks
Buffer's 2026 State of Social Media Engagement report, built on more than 52 million posts, put X's 2025 median engagement rate at about 2.5%, the lowest of the networks it ranked, behind LinkedIn (about 6.2%), Facebook, Instagram, TikTok, Pinterest and Threads. That does not make X useless for traffic, but it does mean each post has to work harder for the same attention, and that X is best treated as one channel in a mix rather than the whole distribution plan.
The audience is smaller and more specific
Pew's 2025 survey of 5,022 US adults found X is used by somewhat fewer adults than WhatsApp (32%) and Reddit (26%), and its fact sheet shows use skews towards college graduates (24%) and households earning $70,000 or more (25 to 26%). X is no longer a mass-reach channel for most brands, but it remains a concentrated one for tech, finance, media, politics and developer audiences, who are often exactly the readers a blog wants. For how it compares with other networks, see Twitter/X vs LinkedIn and Bluesky vs X.
How the 2026 code differs from the 2023 release
X first published its recommendation code in 2023. Hootsuite's summary of that release describes a pipeline that sourced about 1,500 candidate posts per request, ranked them with a machine learning model and ran roughly 5 billion times a day. The 2026 repository keeps the same broad shape (retrieve candidates, rank, filter, mix) but the ranking is now a transformer that predicts a long list of specific actions, including dwell time, link clicks, profile visits and negative feedback. Many of the explicit weights quoted in older "algorithm hacks" posts came from the 2023 code and should not be assumed to apply now.
X is high variance
Buffer's comparison of 1.7 million posts across X, Threads and Bluesky found the median post on all three earned about 4 engagements in 2024, but X's average was 328, against 58 for Threads and 21 for Bluesky, with a far larger spread. Buffer calls X "high risk, high reward": most posts get little, and a few take off. For traffic, that argues for posting steadily and giving every piece several chances rather than betting on one launch-day post.
Why most X posts do not drive traffic
Most traffic-seeking posts fail for the same three reasons. First, they are link-first: a headline, a URL and nothing else, which asks for a click before giving the reader any reason to trust it. Second, they give the algorithm nothing to predict except a click, so they rank below posts that invite replies, dwell time and reposts. Third, they are sporadic: a burst of promotion when an article launches, then silence, so the account never builds the follower base whose in-network posts are not discounted.
The fix is to separate two jobs. The post's job is to earn attention on X: to be worth reading, replying to and reposting on its own. The link's job is to offer more depth to the people who now want it. When the post does its job, the link has an audience. When the post is just a link, it does neither.
The rest of this guide is about doing both jobs well. Our guide to the Twitter/X algorithm in 2026 covers ranking in more depth.
Breakdown 1: by account type
How much traffic X can send you depends first on what kind of account you post from.
| Account type | Median reach per post | Median engagement rate, link posts | Traffic strategy |
|---|---|---|---|
| Regular (no Premium) | Typically under 100 | 0% | Links in replies or profile; build followers with native posts first |
| Premium Basic | A small lift over regular | Not reported separately | Better than regular, but still treat links as secondary |
| Premium | Close to 600 | About 0.25 to 0.3% (Premium tiers overall) | Occasional link posts viable; native formats still lead |
| Premium+ | Over 1,550 | About 0.25 to 0.3% (Premium tiers overall) | The safest tier for link posts |
Source: Buffer's X Premium analysis of 18.8 million posts from 71,000 accounts, August 2024 to August 2025. Figures are medians.
Brand accounts
Brands often post nothing but links to their own content, which is the weakest possible pattern. Mix in native posts that stand alone: tips, data points, product details, answers to common questions. Those posts build the follower base, and followers are where in-network distribution starts.
Founders and executives
Personal accounts with a point of view often out-earn brand accounts on X, because people are more likely to reply to a person than a logo. A founder sharing a lesson from building the product, with a link to the full write-up in a reply, tends to send better-qualified traffic than the brand account posting the same link.
Creators and writers
For newsletter writers and bloggers, X works best as a preview channel: publish the strongest idea from each piece as a native post or thread, and point to the full piece. Consistency matters more than any single post, because followers learn that your posts lead somewhere worth going. Our guide to getting more X followers covers audience building.
Breakdown 2: by post format
Format decides how much attention a post earns before anyone sees the link.
| Format | Median engagement (Buffer 2026) | Regular vs Premium engagement (Buffer, Aug 2025) | Role in a traffic strategy |
|---|---|---|---|
| Text | 3.56% | About 0.40% vs about 0.90% | Best for opinions, hooks and previews of an article |
| Image | 3.40% | About 0.20% vs about 0.42% | Best for data, quotes and visual summaries |
| Video | 2.96% | About 0.25% vs about 0.85% | Best for demos and walkthroughs |
| Link | 2.25% | 0% vs about 0.28% | Weakest; use sparingly or in replies |
Sources: Buffer's 2026 content format analysis and Buffer's links on X analysis. The two use different samples and periods, so compare the ranking, not the absolute numbers. Text leads in both.
Text posts
Text is the native language of X and the top format by median engagement. A strong text post states the most interesting idea from your article in a few lines, so a reader gets real value without clicking and wants more. Use our X character counter to keep it tight.
Image posts
Images carry data, quotes, screenshots and diagrams that make a reader stop. X allows up to 4 photos per post at 5 MB each. A single chart from your article, or four images summarising four points, often earns more clicks on the follow-up link than the link post would have earned alone.
Threads
A thread turns an article into a series of connected posts. It earns dwell time and replies, and the final post is a natural place for the link. Threads take more effort but are one of the most reliable ways to send traffic from X. Our Twitter thread strategy guide covers structure and length.
Video
A short demo or walkthrough works well for product pages and tutorials. Premium accounts' video posts had a median engagement rate around 0.85% in Buffer's data, second only to text.
Breakdown 3: where to put the link
Link placement is the single biggest lever most accounts can pull.
| Placement | Pros | Cons | Best for |
|---|---|---|---|
| In the main post | Simple; one click; shows a link card | Lowest-engagement format; near-zero for regular accounts | Premium accounts, occasional announcements |
| In a reply to your own post | Main post competes as text or image; link still one tap away | "Not foolproof" per Buffer; some readers never open replies | Most accounts, most of the time |
| At the end of a thread | Readers who reach the end are qualified and curious | More effort to write | Long-form articles, guides, research |
| In your bio or pinned post | Persistent; does not affect individual posts | Extra step for readers | Newsletter signup, main offer, lead magnet |
| In an image (as text) | No link penalty | Not clickable; poor experience | Avoid |
Buffer's recommendation is to put links in replies and let the main post earn attention first, while noting the approach is "not foolproof". The code supports the logic: the ranking model predicts actions like replies, dwell and reposts, and a post that earns those signals reaches more people, some of whom will open the reply.
Make the link worth clicking
Wherever the link goes, tell readers exactly what they will get: "Full benchmark with all 12 charts" beats "Read more". Specific promises lift click-through rate, and they set expectations that reduce bounce.
Breakdown 4: by content goal
Different destinations need different post structures. These templates are copy-ready.
| Goal | Post structure | Link line |
|---|---|---|
| Blog article | The most surprising finding in one sentence, then two supporting lines | "The full breakdown, with examples:" |
| Original research | One chart image plus the headline number and what it means | "All 12 charts and the method:" |
| Newsletter signup | A useful tip from this week's issue, given in full | "I send one of these every Tuesday:" |
| Product page | A 20-second demo video of one job the product does | "Try it on your own data:" |
| Event or webinar | One takeaway the session will teach, with date and time | "Save a seat:" |
| Free tool or template | A screenshot of the tool's output | "Free, no signup:" |
Worked example: one article, one week of posts
A fictional analytics startup, Tallyworks, publishes a long article on churn benchmarks. Instead of one link post, the marketer schedules five posts across the week:
- Tuesday 9 a.m.: a text post stating the headline finding, with the link in a reply.
- Wednesday noon: an image post with the article's most striking chart, with the link in a reply.
- Thursday 2 p.m.: a seven-post thread walking through the method, ending with the link.
- Friday morning: a short screen recording of the interactive benchmark tool from the article.
- Next Tuesday: a text post answering the most common question from the week's replies, linking back.
Each post is valuable on its own, each gives the algorithm a different set of actions to predict, and each link is tagged with a different utm_content value, so the marketer can see which format sent the most visitors.
Breakdown 5: by industry and audience
Developer tools and technical products
For technical audiences, code snippets as images, short terminal recordings and threads explaining a technical decision perform well; documentation and changelog links work best in replies.
Finance and investing
Charts and data points dominate. Post the chart natively with a clear takeaway, and link to the full analysis. Keep compliance in mind: the image and text must stand on their own without misleading.
Media and publishing
Publishers are hit hardest by link suppression. The most resilient approach is to publish the key fact or quote natively and accept that X builds brand recall as much as clicks, while newsletter signups capture the readers who stay.
B2B SaaS
Founder accounts with practical lessons outperform brand accounts posting feature links. Use the brand account for product updates with screenshots, and the founder for opinions and stories.
E-commerce
Product images and short videos outperform product links. Use the bio and pinned post for the store link, and keep individual posts native.
For stores with frequent drops or sales, a short countdown series (a teaser image, a close-up detail, then the launch post with the link in a reply) gives followers three reasons to engage before you ask for the click, instead of one.
Creators and newsletters
Treat every post as a free sample. A consistent schedule of genuinely useful posts with a newsletter link in the bio converts better than daily link posts.
Writing the post: hooks that earn the click
The first line decides whether anyone reads the second. These patterns work for traffic because they deliver a specific piece of value and create a reason to want the rest.
| Pattern | Template | Example |
|---|---|---|
| Surprising number | [Number] [unexpected result]. Here's why. | Only 3 of 40 onboarding emails got opened twice. The difference was the subject line. |
| Contrarian claim | Everyone says [belief]. Our data says [opposite]. | Everyone says post daily. Our best month was 8 posts. |
| Before and after | We went from [A] to [B] by changing [one thing]. | We cut support tickets 30% by rewriting one help page. |
| Mistake | The [topic] mistake I made for [time]. | The pricing mistake we made for two years. |
| Checklist preview | [N] checks before you [task]. The first one catches most problems. | 5 checks before you publish a landing page. |
| Question with stakes | Would you rather [A] or [B]? We tested both. | Would you rather have 1,000 followers or 100 subscribers? We tested both. |
The examples are illustrative; replace them with your own numbers. Whatever the pattern, the claim in the hook must be true and supported by the page you link to. A hook that overpromises gets the click once and loses the reader for good.
The body and the call to action
After the hook, give two or three lines of real substance: the method, the key finding, the first step. Then a call to action that names the payoff ("The full checklist, with templates:") rather than "Link below". If the link is in a reply, say so plainly at the end of the main post so readers know to open it.
Setting a realistic traffic goal
It helps to do the maths before judging X. Buffer's medians put a regular account at typically under 100 impressions per post, a Premium account close to 600 and Premium+ over 1,550. If a small share of viewers click, a regular account's median post sends only a handful of visitors, and a Premium account's sends a few more. That is why a traffic strategy on X depends on three multipliers rather than one post: more followers (in-network posts are not discounted), more posts per piece of content (five chances instead of one), and the occasional post that breaks out, which X's high variance makes possible. Track your own baseline for a month before setting targets, and judge X on visitors per hour spent, compared with your other channels.
Repurposing X posts for other text networks
The posts you write to drive traffic from X can do the same job on Threads, Bluesky and LinkedIn with a few changes. Threads allows 500 characters and up to five links per post, so a slightly longer version with the link inline often works there. Bluesky posts are short as well, and LinkedIn's audience expects more context and a professional angle. Write the idea once, then adapt the length, tone and link placement for each network rather than pasting the same text everywhere. Our guide to Bluesky vs X covers how those audiences differ.
When to post for traffic
Timing will not rescue a weak post, but posting when your audience is active gives a strong one a better start.
| Source | Best window | Avoid | Basis |
|---|---|---|---|
| Sprout Social (2026) | Tue to Thu, 12 to 6 p.m. local time | Saturday | Nearly 2 billion engagements, about 307,000 profiles, Nov 2025 to Feb 2026 |
| Buffer (2026) | Tue 9 a.m.; weekday mornings 9 to 11 a.m. | Evenings 6 to 11 p.m.; Saturday | More than 8.7 million posts |
Sources: Sprout Social and Buffer. The two disagree on morning versus afternoon, which usually means your own audience's data matters more than either. Test both windows for a month. For traffic specifically, also consider when your audience can actually read an article: a Tuesday morning post may get the reaction, but a link in a reply may be opened at lunch. Our best time to post on X guide has day-by-day detail.
Images and link previews that earn clicks
Native images
Use 16:9 (1600 x 900) or square (1080 x 1080) images, per Sprout Social's size guide, and keep text large enough to read at thumbnail size, since multi-image posts display as a grid. A chart, a screenshot of a key paragraph, or a simple framework diagram gives the reader a reason to open the full article. Our X image resizer handles dimensions, and images on Twitter/X covers what works creatively.
Link cards
When a link does appear in a post, X builds a preview card from the page's metadata. The Open Graph protocol defines four required properties, og:title, og:type, og:image and og:url, plus optional og:image:width, og:image:height and og:image:alt. Make sure every article has a strong og:image at a wide aspect ratio and a title written for social, not just for search. A blurry or missing preview image costs clicks.
Your profile as a traffic hub
Every post that does well sends some readers to your profile, and the X ranking model even treats profile clicks and profile visit time as predicted actions in its open-source code. Make the profile convert those visits.
- Bio: one line on what you post about and who it is for, plus one link to your highest-value destination (usually a newsletter or a resource hub), tagged with UTMs.
- Pinned post: your best-performing native post or thread on your core topic, ending with the link. It is the first thing a curious visitor reads.
- Header image: a 1500 x 500 header, per Sprout Social's size guide, that states what the account offers. The profile photo overlaps the lower left, so keep text right of centre.
- Consistency: a visitor who scrolls your recent posts should see a steady stream of useful posts, not a wall of links.
Replies as a traffic source
Thoughtful replies to larger accounts in your niche put your name in front of their audience, and some of those readers visit your profile. The key word is thoughtful: add a data point, a counterexample or a short how-to, not "Great post!". Do not drop links in other people's replies; it reads as spam and invites the negative signals (mute, block, report, "not interested") that X's model subtracts from a post's score. Let the reply earn the profile visit, and let the profile carry the link.
The same applies to your own posts. Replying to people who respond to you keeps the conversation going, and replies are among the actions X's ranking model predicts. Block 15 minutes after each scheduled post goes out to answer questions.
Launch days without the link dump
Launches tempt teams into posting the same link five times in a day. X's ranking applies an author-diversity discount, so when several of your posts compete for the same reader's feed, each additional one is scored lower. A launch-day link dump mostly competes with itself. Spread launch posts across a week instead: announcement with a demo video on day one, a customer or data story on day two, a thread on how it works on day three, and an answer to the most common question on day five. Each post brings a different reason to click.
Alt text on every image
Add alt text to every image you post on X. It makes charts and screenshots usable for screen-reader users, and it forces you to state what the image shows, which often reveals whether the image actually supports the click you are asking for. If an image contains the key number, put that number in the alt text and in the post text too.
Measuring traffic from X
If you cannot see which posts sent visitors, you cannot improve. Two layers of measurement cover it.
UTM parameters
Google's documentation says that when you add parameters to a URL, "you should always use utm_source, utm_medium, and utm_campaign", and that missing parameters show as "(not set)" in reports. Add utm_content to tell posts apart. Because X counts every URL as 23 characters, long UTM-tagged links cost nothing extra in the post. Our UTM builder creates tagged links, and the UTM parameters glossary entry explains each one.
| Parameter | Example value | What it tells you |
|---|---|---|
| utm_source | x | Traffic came from X |
| utm_medium | social | Organic social, not paid |
| utm_campaign | churn-benchmark-2026 | Which piece of content or campaign |
| utm_content | thread or chart-image or reply-link | Which post format or placement sent the click |
On-platform and site metrics
On X, watch link clicks, profile visits and replies. In your analytics, watch sessions, engaged sessions and conversions by utm_content. Compare formats monthly: if threads send three times the visitors of single posts, write more threads. Our guide to X metrics and KPIs covers which numbers matter.
How to drive traffic from X: step by step
- Decide whether Premium is worth it. If X traffic matters to your business, Buffer's reach data makes a strong case for at least the Premium tier.
- Fix the destination first. Check every page you will link to has a strong Open Graph image and title, and loads fast on mobile.
- Turn each article into several native posts. One headline finding, one chart, one thread, one demo, one follow-up answer.
- Write each post to stand alone. A reader who never clicks should still learn something.
- Choose the link placement per post. Reply link by default; end of thread for long pieces; main post only on Premium and only occasionally.
- Tag every link. utm_source, utm_medium and utm_campaign always; utm_content per post.
- Schedule across the week. Spread posts over weekday windows rather than publishing all on launch day.
- Be present after posting. Reply to replies in the first hour; the ranking model predicts replies and conversation.
- Measure after four weeks. Compare visitors and conversions by utm_content and by day and time.
- Double down. Write more of the formats and topics that sent engaged visitors, and drop the rest.
Common mistakes and how to fix them
- Link-only posts. Fix: give the value in the post; offer the link as more depth.
- Posting links from a regular account and expecting reach. Fix: links in replies, or consider Premium.
- "Read more" calls to action. Fix: say exactly what the click delivers.
- One post per article. Fix: five posts across a week, in different formats.
- Untagged links. Fix: UTM parameters on every link, with utm_content per post.
- Missing or broken link previews. Fix: set og:image and og:title on every page.
- Posting everything on launch day. Fix: schedule across the week so each post has room.
- Ignoring replies. Fix: answer in the first hour; replies are a predicted action.
- Text in images that should be clickable. Fix: never put a URL only inside an image.
- Posting at night. Fix: weekday mornings and early afternoons in your audience's time zone.
- Inconsistent posting. Fix: a steady weekly schedule builds the follower base that sees in-network posts.
- Judging X by one post. Fix: measure a month of posts by format and placement before deciding.
How to do this with Postprism
Postprism is a social media scheduler that publishes to 6 networks, including X, from one content calendar. For a traffic strategy, that means:
- A week of posts per article. Plan the headline post, chart post and follow-ups in the content calendar and drag them into the best weekday slots.
- Images on X. Schedule X posts with up to 4 images, so your chart or screenshot posts go out on time.
- Time zones. Choose the time zone you are scheduling for, so 9 a.m. means 9 a.m. for your audience.
- Bulk scheduling. Load a month of posts, each with its own UTM-tagged link, from a CSV with bulk scheduling.
- Per-network versions. Publish the same idea to LinkedIn and Bluesky with a version written for each, from the same post.
- Measurement. Compare results in analytics and export to CSV, alongside your UTM data.
The X scheduler page covers what Postprism publishes to X.
Schedule Your Twitter/X Posts in Advance
Plan a week of traffic-driving posts for every article, schedule them for your audience's best hours, and publish to X and 8 other networks from one calendar. 7-day free trial.
Frequently Asked Questions
Does X suppress posts with external links?
X has not published a link penalty, but the data shows links are the weakest format. Buffer found link posts had a 2.25% median engagement rate against 3.56% for text, and regular accounts' link posts have had a 0% median engagement rate since March 2025. Premium accounts still get some reach on links.
Should I put the link in a reply instead of the main post?
For most accounts, yes. It lets the main post compete as a text or image post while keeping the link one tap away. Buffer recommends it while noting it is not foolproof. For long articles, a thread that ends with the link works well too.
Is X Premium worth it for driving traffic?
If X traffic matters to your business, probably. Buffer's analysis found Premium accounts get roughly 10x the median reach of regular accounts, and their link posts still earn engagement where regular accounts' do not.
How many characters does a link take on X?
Every valid URL counts as 23 characters regardless of its actual length, because X wraps it with its t.co shortener. Long UTM-tagged links cost no more than short ones.
When is the best time to post on X for traffic?
Sprout Social's 2026 data points to Tuesday to Thursday, 12 to 6 p.m. local time; Buffer found 9 a.m. on Tuesday was the best single slot. Start with weekday mornings and early afternoons in your audience's time zone, and test.
Do threads drive more traffic than single posts?
Often, for long-form content. A thread earns dwell time and replies, and readers who reach the final post with the link are already interested. Measure it yourself with a different utm_content value for thread links and single-post links.
Should I use link shorteners on X?
There is no character benefit, because X counts every link as 23 characters. Use your own domain with UTM parameters so readers can see where they are going and you can track the click.
How do I know how much traffic X sends my site?
Tag every link with utm_source, utm_medium and utm_campaign (plus utm_content per post) and check sessions and conversions in your analytics. Compare against link clicks in X's own analytics to see how many clicks become real visits.




