Semrush

Semrush can put traditional search research, technical auditing, rank tracking and AI-search monitoring under one brand. It does not put every capability under one simple subscription. SEO Classic, AI Visibility, Semrush One, Content and other toolkits have different plans, limits, users and add-ons. Choosing “Semrush” before choosing the exact job is therefore the fastest route to an expensive, incomplete setup.

> Distinctive strength: A search team can investigate a domain, audit it, track its positions and compare sampled AI-answer visibility within the same established project ecosystem instead of rebuilding domain and competitor context in separate tools. > > Where it stops being an advantage: The shared brand does not mean one entitlement: SEO Classic, AI Visibility, Semrush One, Content and add-ons have different users, meters and prices.

BenPicks has not run a controlled accuracy or productivity test of Semrush. We do not claim its keyword estimates, audit findings, AI Visibility Score or content recommendations outperform alternatives. The decision this page supports is more useful: which Semrush package exposes the required workflow, what will the complete workload cost, what do its metrics really mean and which legal or operational boundaries should stop the purchase?

The decision in 60 seconds

Which Semrush product are you actually buying?

The current product architecture separates several buying jobs:

SEO Classic contains the familiar keyword, competitor, backlink, Site Audit and Position Tracking workflows. Its Pro, Guru and Business plans mainly change monitored websites, tracked keywords, crawl capacity, historical data, content features and API access.

AI Visibility Toolkit adds large-dataset AI research, Brand Performance, competitor analysis, Prompt Tracking and AI Search Site Audit. Its standalone subscription is centered on one folder/domain and a small daily prompt set.

Semrush One bundles SEO and AI Visibility under Starter, Pro+ and Advanced. It is not merely a discount code for two identical standalone products: its monitored websites, keywords, prompts, content capabilities and API access rise by tier.

Content Toolkit is a separate content creation/optimization path. Enterprise AIO is the custom path for global scale, integrations and services.

Write the required toolkits before comparing prices. Otherwise, a buyer can purchase SEO Classic and discover that the AI workflow is separate—or buy Semrush One when a narrower $99 monitoring job was sufficient.

Which Semrush plan combines the SEO and AI-search work required?

Choose by workflow boundaries, not by the Pro/Guru/Business names you remember. The public USD surfaces checked on 27 August 2026 showed:

The annual figures are monthly equivalents of an annual billing choice, not month-to-month prices. Semrush labels some Semrush One pricing introductory. Currency, tax, current account offers and renewal terms require checkout confirmation. Because the commercial facts combine multiple toolkits, meters and commitments, BenPicks omits `Offer` schema rather than publishing a deceptive single price.

How much does Semrush cost after domains, prompts and users?

The answer can be far above the starting price because AI Visibility scales by monitored domain/location, prompt capacity and licensed user.

Standalone AI Visibility currently publishes:

• one folder and one domain for Brand Performance;
• 25 prompts for Prompt Tracking;
• 300 AI Analysis report queries per day;
• 1,000 Prompt Research queries per day;
• ten CSV exports per day;
• $99/month for each additional Brand Performance domain or location;
• $60/month for 50 additional prompts;
• another $99 license for each corporate subuser needing the toolkit.

Representative arithmetic:

1. SEO-only individual: $139/month. SEO Classic Pro is the smallest current paid SEO tier. This excludes AI Visibility, extra users, specialized reports and other toolkits.

2. Separate SEO + AI subscriptions: $238/month. Pro ($139) plus AI Visibility ($99) totals $238 before tax. Semrush One Starter is $199. The $39 difference does not automatically make the bundle better: compare included websites, prompts, keywords, features and future tier needs.

3. Two-domain AI monitoring: $198/month. Base AI Visibility ($99) plus one additional domain ($99). Add 50 prompts and the subtotal becomes $258. Add one extra AI Visibility user license at $99 and it becomes $357.

4. Growing bundled team: at least $389/month. Semrush One Pro+ ($299) plus two users at the general “starting at $45” price equals at least $389. Toolkit- and tier-specific user prices must be confirmed before budgeting.

5. Advanced bundle with reporting and another market: $668/month. Advanced ($549) + Pro Report ($20) + one additional AI domain or location ($99) equals $668 before users and tax.

Use this formula:

monthly cost = toolkit/bundle + additional domains + additional locations + prompt capacity + licensed users + report products + API/other apps + analyst/editor time

Cancellation is not an instant escape from a wrong annual decision. Semrush's online policy says cancellation takes effect at the end of the committed term. Its one-time seven-day refund route is limited to qualifying initial annual subscriptions or initial add-on purchases made directly by card; month-to-month subscriptions are excluded. Eligible trials bill automatically unless cancelled correctly before they end.

What does Semrush's AI Visibility data actually measure?

It is a set of modeled and sampled datasets, not one universal measurement of people seeing a brand in AI. Semrush documents at least three distinct systems.

1. Prompt Database

Visibility Overview, Competitor Research and Prompt Research use a database Semrush describes as more than 317 million prompts/responses across ChatGPT, Gemini, Google AI Overviews and AI Mode, covering 117 regional databases with rolling daily updates.

Semrush says prompt inputs are sourced from AI-search clickstream data and Google's keyword data, then grouped into topics, deduplicated and simplified while preserving intent. This is broad research coverage. It is not every private prompt and not a stable panel of identical users and models.

2. Brand Performance

For a selected domain and location, Semrush generates branded and non-branded synthetic prompts and checks several platforms. These reports refresh weekly. They can describe the narrative around a brand, but the prompt set is constructed by Semrush—not an independently observed list of every buyer question.

3. Prompt Tracking

The buyer selects prompts, platform and location, and Semrush queries them daily through Position Tracking. This is the most controlled monitoring surface, but AI responses still vary. One daily observation cannot prove that an underlying answer is stable or that a page edit caused the change.

How to interpret the headline metrics

AI Visibility Score runs from 0 to 100 and combines topic coverage with mention consistency. It is normalized within Semrush's topic and competitive model. It is not a conventional search rank.

AI Volume groups semantically related prompts and combines third-party interaction data with Semrush machine-learning models. Monthly Audience estimates query audience across topics where a brand appears. Neither is observed unique visitors, referral traffic or conversions.

Semrush itself states that AI answers are changing and personalized and that no platform can provide exact AI-visibility numbers. Treat the metrics as directional signals for prioritization and controlled monitoring. Validate actual traffic with first-party analytics/Search Console and actual sales questions with customer evidence.

Can Semrush replace a specialist content or technical SEO tool?

Only if it passes the specialist workflow on your material—not because the bundle includes a similarly named feature.

On technical SEO, Site Audit can crawl a domain, subdomain or subfolder. Current documentation lists monthly capacities of 100,000 pages for Pro, 300,000 for Guru and one million for Business; per-audit limits are 20,000, 20,000 and 100,000 respectively. Scope, authentication, JavaScript and crawl settings influence what the crawler actually sees.

A large allowance does not prove issue quality. Test known defects, false positives, prioritization and the time required to reach an actionable fix list. Site Health is a Semrush scoring construct, not proof that a site is technically sound or will rank.

On content, Semrush documents optimization against factors associated with SEO/AI citations, while some Semrush One levels include SEO Writing Assistant and Topic Research. The correct comparison against Surfer, Frase or Clearscope is not “number of recommendations.” Use one real brief and record:

• important concepts detected or missed;
• unsupported/factually wrong suggestions;
• duplicated competitor framing;
• editor correction time;
• final differentiation and readability;
• whether a recommendation produced a publishable change.

A score improvement is not the success metric. The deliverable is.

Semrush also exposes many exports and scheduled reports. That supports hand-off, but it is not one universal CMS publication workflow. Site Audit results can export crawled pages/issues to spreadsheet/PDF formats; Prompt Tracking supports Excel, CSV, Google Sheets, PDF and scheduled delivery. Confirm daily export limits and client-facing cleanup before assuming reporting is solved.

Are the workspace, API and data terms acceptable for client work?

They can be, but the exact role, toolkit license, input type and downstream use must pass policy review.

Semrush documents Owner, Admin, User and Guest roles plus folder sharing. Users share subscription limits, and Full Sharing can expose team assets broadly. Separately purchased AI Visibility access can require an additional $99 license for each subuser. Test real accounts: a role label alone does not prove client isolation, least privilege or the expected toolkit access.

API access appears at the Business/Advanced end of current pricing. The terms restrict usage to ten inquiries per second per IP, ten simultaneous inquiries per authorized user and no caching of received information beyond one month without written permission. A data warehouse or client product built on Semrush data needs a legal/technical design, not just an API key.

For generative AI functions, Semrush says neither it nor its third-party providers train models on User Input. However, the same AI terms allow Semrush to use, store, modify, edit, delete and transfer inputs to third-party AI providers to provide and improve the service. “No training” is not “no storage,” “no transfer” or “zero retention.”

The terms say users retain input rights and receive a broad license to output only to the extent Semrush has rights. Output may be inaccurate, biased, incomplete, non-unique or unsuitable; users remain responsible for it.

Most importantly, general terms say the services are not designed to process sensitive information, explicitly including protected health information, nonpublic financial information and protected special-category data. If the workflow requires those inputs, do not attempt to solve the problem with a workspace setting.

Semrush publishes a DPA, security information and a subprocessor list including infrastructure and analytics providers across named locations. Those are procurement inputs. They do not make a customer's configuration, permissions, inputs, exports or integrations compliant automatically.

When should you exclude Semrush?

Choose another route or postpone purchase when:

• the team cannot name the required toolkit before checkout;
• the decision requires actual AI audience/referral measurement rather than modeled visibility;
• domain, location, prompt and user multiplication makes the AI setup uneconomic;
• sensitive data must enter the service;
• a separate license per teammate breaks the collaboration model;
• API rate/caching restrictions conflict with the intended product or warehouse;
• a specialist content/audit tool wins a declared representative test;
• a public contractual uptime SLA is mandatory and cannot be obtained in writing;
• no person owns the action after a gap, issue or visibility change is reported.

Compare Ahrefs when web-index research plus macro AI discovery is central, Surfer SEO when content optimization is the primary production job, or browse the AI SEO category to separate monitoring, research and editing. Use the comparison tool only after freezing the required workload and hard gates.

A product-specific Semrush evaluation protocol

1. Write the required toolkit list: keyword/competitor research, audit, rankings, AI discovery, custom prompts, content, reports and API.
2. Freeze one owned site, one competitor, one market, 20 commercial queries and 20 known technical defects.
3. Run Site Audit with declared scope and page limit. Record known-issue recall, false-positive review, crawl duration, pages consumed and operator time.
4. Track 50 representative keywords. Compare results with Search Console and controlled direct checks; retain disagreements.
5. Label every AI metric as observed, modeled or normalized from Semrush's own methodology before using it in a report.
6. Configure ten custom prompts with exact platform and location. Preserve daily answers, mentions, citations, model context and variance.
7. Compare Brand Performance's generated prompts with real sales, support and research questions. Count missing commercial intent and irrelevant prompts.
8. Trace ten cited pages or competitor gaps to actions with owners. Reject a report whose only outcome is another dashboard.
9. Test the content workflow on one real brief against the current specialist/editorial process. Count factual errors, missing coverage, revisions and final operator time.
10. Export the exact client reports. Measure limits, format, cleanup, scheduling and access control.
11. Create test users with intended roles and licenses. Verify folder visibility, history, toolkit access, limit allocation and offboarding.
12. If API is required, test representative endpoints within request and caching restrictions; record errors, latency and projected monthly data cost.
13. Review AI input processing, sensitive-data prohibition, DPA and subprocessors before adding confidential/client information.
14. Recalculate complete cost from bundle/toolkits, domains, locations, prompts, users, reports, other apps and labor.
15. Apply thresholds declared before testing: maximum cost, minimum known-issue recall, maximum cleanup time, acceptable data boundary and minimum actionable findings.

BenPicks has not executed this protocol. It is a reader test plan, not an invented claim of hands-on use, ranking gain or AI-visibility improvement.

What remains unknown

The retained public pricing, legal, security, support and Knowledge Base sources do not establish one generally applicable self-serve uptime SLA with a measurement definition and service-credit schedule. Customer Support Terms promise commercially reasonable efforts but explicitly do not guarantee resolution of every problem.

If uptime is a hard gate, request the exact percentage, measurement window, excluded events, claim procedure, response targets and remedy in the signed order. Until that exists, the field remains Unknown.

Verdict

Semrush is strongest when a team genuinely needs traditional SEO and AI-search evidence in a connected workflow and will act on both. Its current package breadth can be efficient; it can also hide duplicated capacity and domain/user costs.

Start by selecting the job and toolkit. Compare Semrush One against separate subscriptions using actual websites, keywords, prompts, locations and users. Treat AI scores and audience as modeled directional indicators, not traffic. Keep sensitive information out. If a representative audit/content/monitoring test does not reduce decision time or improve action quality enough to justify the full bill, the bundle has not earned its place.

Official sources

• SEO Classic pricing ↗ — plans, annual display, limits and add-ons.
• Semrush One pricing ↗ — bundle tiers, websites, keywords and prompts.
• Semrush One overview ↗ — toolkit composition and evaluation path.
• AI Visibility Toolkit ↗ — standalone price, domains, prompts, users and limits.
• AI Visibility data methodology ↗ — prompt sources, refresh cadence, scoring and modeled volume.
• AI Visibility features ↗ — workflow boundaries and free access.
• Prompt Tracking ↗ — metrics, exports and scheduling.
• Site Audit configuration ↗ — scope and crawl limits.
• User Management ↗ and Folders ↗ — roles, sharing and monitored-site limits.
• Terms of Service ↗ — API, sensitive-data and subscription boundaries.
• AI Functionality Terms ↗ — input processing, no-training statement and output license.
• Cancellation and Refund Policy ↗ — cancellation, trial billing and narrow refund eligibility.
• DPA ↗, Security ↗ and Subprocessors ↗ — procurement evidence.
• Customer Support Terms ↗ — support scope and non-guarantee.