The 10-Point New Site SEO Checklist for 2026

You're hours from launch. Design is signed off, the CMS is populated, staging looks clean, and everyone wants to push the button. Then the SEO questions start. Are the right pages mapped to the right queries? Did anyone leave a noindex tag on a money page? Will Google understand the site structure on day one? And in 2026, one more question matters just as much. Will AI answer engines find, interpret, and cite the site at all?

That's where most launch plans still fall short. They handle the usual mechanics like titles, redirects, and sitemaps, but they don't account for the way people now discover brands through ChatGPT, Perplexity, Gemini, and Google AI Overviews. According to the research cited in Netcode's roundup on SEO checklists for new websites, AI Overviews appear in 15-20% of Google searches, up from 7% in 2024, and Perplexity handles 50M+ queries daily. If your launch process ignores that shift, you're optimizing for only part of search.

That doesn't mean traditional SEO matters less. It means the bar is higher. You still need clean technical implementation, keyword mapping, Search Console setup, schema, mobile-first performance, and crawl controls. You also need content that answers questions directly, a structure AI systems can extract from, and monitoring that tells you whether your brand is appearing in machine-generated answers.

This new site seo checklist is the launch plan I'd use for a serious site build today. It's built for practitioners, not theory. If you're launching a startup site, rebuilding an enterprise property, or preparing a client migration, treat this as go-live insurance and growth planning in one document. If you're handling the wider launch process too, this guide for founders launching products is a useful companion.

1. Phase 1 Pre-Launch Keyword and Content Strategy for AI

A site launch goes sideways fast when the team approves templates before it knows which questions the site needs to answer. I have seen clean builds go live with polished design, then stall because the content plan was built around broad category terms instead of the actual language buyers, evaluators, and AI systems use to retrieve answers.

Start with the commercial core. Define 5 to 15 cornerstone keywords tied to revenue-driving services, products, or categories, then verify search demand and SERP intent in Google Keyword Planner, Ahrefs, or Semrush. That keyword set gives the site its primary architecture, but it should not be the whole brief. For a new site in 2026, the job is broader. You need targets that can rank in search and also feed answer engines with clear, quotable information.

Map intent before page creation starts

Keyword mapping is not a spreadsheet exercise you clean up later. It decides which pages exist, what each page is responsible for, and where overlap will cause problems.

Assign one primary intent to one page. If a service page, comparison page, and blog post could all rank for the same query, choose the page that should win and rewrite the others around a different angle. That trade-off matters early, especially on new domains with limited authority. Spreading similar intent across multiple URLs usually weakens all of them.

A stronger pre-launch map includes both classic search behavior and answer-engine behavior:

• Primary keyword: The main commercial term for the page.
• Intent type: Transactional, comparative, informational, support, or local.
• Question variants: The exact phrasing prospects use in sales calls, internal search logs, Reddit threads, and SERP People Also Ask results.
• Citation target: The question an AI system should be able to answer from this page in one extractable passage.
• Page owner: The single URL responsible for that topic.

That last field matters more than teams expect.

For example, a B2B payroll SaaS company might target "payroll migration services" with a money page, then support it with separate assets for "how to switch payroll providers," "[competitor] vs [brand]," and "payroll migration checklist." Those are related topics, but they serve different intents. Treated as one blob, they compete with each other. Mapped correctly, they build topical coverage without cannibalization.

Add prompt research to the keyword brief

Classic keyword tools still matter. They just miss part of the discovery path.

Before content production starts, test the topic set in ChatGPT, Perplexity, Gemini, and Google AI Overviews. Look at which brands get cited, what subtopics appear in generated answers, and where the model gives thin or generic responses. That gap analysis often exposes easier openings than the head term itself. If every competitor has a page on "benefits of endpoint security" but nobody clearly answers "when endpoint security is not enough" or "how endpoint security works in hybrid environments," that is a practical content opportunity.

I use prompt testing to shape briefs, not replace keyword data. The two work together. Search tools show demand. AI testing shows extractability, citation patterns, and whether your future page is likely to be summarized accurately. If your team is building this into process, a practical framework is to pair the keyword map with a generative engine optimization workflow before any page outlines are approved.

Multi-location and multi-service sites need even tighter planning. The problem is not just scale. It is duplicate intent across city pages, service pages, and hybrid local-service templates. A business with many locations and several service lines can end up managing hundreds of keyword-page combinations quickly, which is exactly why pre-launch mapping needs rules for URL ownership, template variation, and internal linking from the start.

The standard new site SEO checklist stops at "pick keywords and create pages." A better launch plan asks a harder question. Which page should rank, which page should be cited, and which phrasing should an AI system lift when a prospect asks for help?

2. Phase 1 Pre-Launch Content Optimized for AI Citation

A new site doesn't just need publishable content. It needs citable content. That's a different standard.

AI systems tend to favor pages that answer the main question quickly, define terms clearly, and present information in a structure that's easy to extract. If your service page hides the answer under a vague brand paragraph and three design-heavy blocks, it may look polished but still perform poorly in both search and AI discovery.

Write the answer early

I like the first paragraph to answer the H1 directly. If the page is “What is cloud cost optimization,” the opening should define it plainly, mention the business value, and set up the explanation. Don't make the reader or the model hunt for the point.

For example, a payroll SaaS page targeting “how to switch payroll providers” should open with a concise summary of the migration process, the main risks, and the conditions that make switching worth it. Then the page can expand into timeline, data transfer, compliance checks, pricing factors, and FAQ.

Here's the formatting pattern that tends to work well:

• Direct answer first: Put a concise response near the top of the page.
• Scannable hierarchy: Use descriptive H2s and H3s that mirror user questions.
• Defined entities: Name the product, service, audience, and use case clearly.
• Visible expertise signals: Show who wrote or reviewed the content and why they're credible.

A lot of teams overcomplicate this. They think AI optimization means sounding robotic. It doesn't. It means being easy to quote accurately.

Use structure that machines can lift cleanly

Schema helps, but page writing matters just as much. FAQ blocks, short definitions, comparison sections, and step lists give answer engines clearer extraction points. The technical side of this overlaps heavily with Generative Engine Optimization principles, especially when you want content to surface in answer-led experiences instead of only traditional listings.

A practical example. If you're launching a cybersecurity consultancy site, don't publish only a broad “managed security services” page. Also publish a sharply structured page answering “what's the difference between MDR and MSSP,” with a summary table-style explanation in prose, implementation considerations, and a clear recommendation by company stage.

Clear pages get reused. Clever pages often get skipped.

One caution here. Don't invent statistics just to make a page look more authoritative. If you don't have verified data, say it qualitatively. Explain patterns, trade-offs, and operational realities in plain language. That's more trustworthy than fake precision, and it ages better.

3. Phase 2 Launch-Day Technical SEO and AI Crawlability

Launch day is where avoidable mistakes become expensive. The site may look complete in staging and still be invisible to crawlers, broken on mobile, or inaccessible to AI bots because of one bad rule.

The technical foundation still decides whether your content has a fair chance to rank, get indexed, and be cited. Before migration, key elements need to be in place: robots.txt, XML sitemap submission, verification that noindex and nofollow tags aren't blocking important pages, structured data markup, and mobile-first responsive design, all of which are emphasized in Cronyx Digital's site launch checklist.

Use a visual checklist for the final pass.

The launch-day crawl test

I'd run three checks before DNS changes or the production switch:

• Browser check: Spot-check templates on desktop and mobile.
• Crawler check: Crawl the staging or pre-live environment with Screaming Frog.
• Render check: Confirm key content is present in rendered HTML, not hidden behind client-side interactions.

Ensuring AI-facing visibility depends on the same baseline accessibility. If product specs, pricing context, or key service descriptions only appear after heavy JavaScript execution, some systems won't interpret the page cleanly.

A common real-world problem is the faceted ecommerce menu that looks harmless in design review but creates crawl traps or hides critical category text. Another is the comparison page built in tabs, where the most useful content never appears in initial render. Both can weaken discovery.

Performance is part of crawlability

Core Web Vitals deserve launch-day attention, not post-launch cleanup. In 2026, only 42% of newly indexed sites achieve “Good” mobile scores across LCP, INP, and CLS, based on Google Search Console aggregate data from over 1 million newly indexed sites referenced in SiteGround's 2026 SEO checklist article. The same source notes that LCP fails most often because hero images exceed 100KB without proper compression or lazy-loading.

That's why I push teams to fix obvious performance debt before launch. Compress hero media, preload critical fonts, defer non-critical JavaScript, and make sure templates don't shift on load. If a publisher can keep pages loading cleanly and quickly, both users and crawlers benefit.

For a walkthrough format, this launch video is useful before final QA:

4. Phase 2 Launch-Day LLMs.txt Implementation

Most SEO teams know what to do with robots.txt. Far fewer have a policy for AI systems. That gap is growing more obvious with every launch.

An LLMs.txt file gives you a way to declare how AI systems should understand your site, your brand name, and your preferred citation context. It's still an emerging standard, but I see it as low effort and worth doing at launch, especially when brand representation matters.

What to include in LLMs.txt

Think of this file as guidance, not magic. It won't force every model to behave exactly as you want, but it gives your site a clean statement of intent.

A practical implementation usually includes:

• Preferred brand naming: State how the brand should be cited.
• Primary expertise areas: Clarify the topics your site is authoritative on.
• Content distinctions: Separate factual resources from opinion or commentary if relevant.
• Contact pathway: Provide a contact for AI partnership or content questions when appropriate.

A straightforward example would be a healthcare software company specifying that the brand should be cited by company name, not domain name, and identifying its strongest subject areas as patient intake, scheduling, and claims workflows. A publisher might also identify editorial sections versus sponsored or opinion-led sections.

Keep it simple and publish it at the root

Don't overengineer the first version. A clean file in the root directory is better than debating wording for three weeks and launching without one. The LLMs.txt generator from LLMrefs is a practical way to create a standards-aligned starting point quickly.

A good launch habit is to publish LLMs.txt at the same time you publish robots.txt, not as a “we'll come back to it” task.

I'd also make sure legal, editorial, and SEO agree on naming conventions before it goes live. If your company is commonly referred to three different ways across your site, AI systems can reflect that inconsistency back to users. This file won't solve a messy brand system on its own, but it helps reinforce one clear version.

5. Phase 3 Post-Launch Brand Mention and SOV Monitoring

The first week after launch tells you whether the site is merely live or visible. Rankings matter, indexed pages matter, and brand mentions inside AI answers matter too. If you're not measuring those things from day one, you're operating on vibes.

I'd set up a post-launch dashboard immediately. Traditional metrics should include Search Console coverage, sitemap status, indexed pages, top landing pages, and event tracking. Cronyx's launch guidance also stresses documenting keyword rankings, organic traffic levels, top-performing pages, conversion rates, and key events before launch so you have a real baseline to compare against later.

Establish the baseline while it's still clean

For AI visibility, a new site often starts near zero. That's normal. What matters is that you capture the starting point before your team starts publishing more pages, doing outreach, or changing templates.

A useful operating setup in LLMrefs is to track your main commercial topics, your brand name, and a few competitor terms right away. That lets the team see whether the site is being cited, ignored, or misattributed across answer engines.

I like to monitor three views in the first month:

• Brand mention presence: Are you appearing at all in answer-engine responses?
• Citation context: Which pages get cited, and for what questions?
• Competitor overlap: Which brands show up where you expected to appear?

A real scenario. A SaaS team launches with strong feature pages but notices the site gets no mentions for comparison prompts. The issue often isn't authority alone. It's usually that the site hasn't published comparison content in a structure AI systems can reuse.

Don't collapse all visibility into one KPI

Share of voice is useful, but don't let it hide what's really happening. A site can gain mentions on informational prompts and still miss the commercial prompts that influence pipeline. Likewise, a spike in one AI model doesn't mean the broader ecosystem has moved.

That's why I prefer segmented monitoring. Track by topic group and by platform. LLMrefs is especially useful here because it helps teams benchmark mention and citation patterns in a format that non-SEO stakeholders can understand.

6. Phase 3 Post-Launch Content Gap Analysis

Competitor analysis gets much sharper once you stop looking only at rankings and start looking at citations. Search results tell you who ranks. Answer engines also reveal who gets reused as a source.

That difference matters for a new site. You may not outrank established domains quickly, but you can still identify where cited content is thin, outdated, biased, or incomplete. Those are the openings worth chasing.

Inspect the cited pages, not just the winning domains

If a competitor keeps appearing for “how to implement single sign-on,” don't stop at noting their domain. Open the page and inspect the structure. Does it define the term clearly? Does it include implementation steps? Does it explain failure points? Does it offer a stronger answer than your draft page?

Teams make real gains by prioritizing pages that have already proven citable in-market, instead of publishing broad content calendars.

A practical weekly review process looks like this:

• Track priority prompts: Pull your most important informational and commercial queries.
• Inspect cited sources: Identify which pages are repeatedly referenced.
• Classify the angle: Is the cited content educational, comparative, transactional, or glossary-style?
• Build the better asset: Improve on freshness, clarity, examples, and author credibility.

For instance, a developer-tools company might discover that competitors dominate “how to implement webhook retries” while nobody owns “when webhook retries create duplicate actions.” That second topic may be less obvious, but it can attract strong citation behavior because it answers a real operational concern.

Look for formatting gaps too

Content gaps aren't always topical. Sometimes the topic exists on your site, but the format is weak. I've seen pages miss citation opportunities because they buried the answer, lacked subheadings, or had no visible author information.

Sometimes the gap isn't “we need a page.” It's “we need a better answer.”

That's one reason I like using LLMrefs after launch. It helps surface not just whether a competitor is winning, but where their winning pages create a pattern you can improve on. That's more useful than building content from a blank sheet.

7. Phase 3 Post-Launch Authority and E-E-A-T Signal Building

New domains don't get the benefit of the doubt. If the site is fresh, thin on reputation, and vague about who wrote the content, both users and machines will hesitate to trust it.

Post-launch, I'd push authority building on two fronts at once. First, make the site itself more trustworthy. Second, earn external signals that reinforce that trust.

Build trust on the site before chasing mentions

Start with visible authorship. Every substantial article should have a real author or reviewer, and that person should have a detailed bio page. If you have subject matter experts inside the company, use them. If not, get content reviewed by practitioners who can speak credibly about the topic.

The trust layer should also include:

• Detailed About page: Explain who runs the company and what they do.
• Clear contact information: Show legitimate ways to reach the business.
• Policy pages: Privacy, terms, and any necessary trust or compliance pages.
• Proof of experience: Use case studies, implementation stories, reviews, or customer examples where appropriate.

A healthcare startup launching educational pages about claims workflows shouldn't hide the people behind the content. It should identify the operator, consultant, or product specialist who reviewed those pages. A fintech site discussing compliance-heavy topics should be explicit about who contributes and what their area of expertise is.

Earn authority outside the site too

This part takes longer, but it compounds. Relevant citations, coverage, and links from respected industry sites help search engines and AI systems build confidence in your brand associations over time.

The mistake I see most often is teams chasing volume too early. They submit to low-quality directories, buy weak placements, and call it authority building. That rarely helps much. One respected mention in a legitimate niche publication can do more for perceived trust than a pile of forgettable placements.

For a new site seo checklist, this is the least “launchy” step but one of the most important. The technical work gets you into the game. Authority is what makes the site durable.

8. Phase 3 Post-Launch Backlink Strategy for AI Training Data

Not every backlink carries the same strategic value now. If you're trying to improve visibility in answer engines, relevance and reputation matter more than sheer count.

I'd rather earn one link from a respected industry publication, a serious trade association, or a trusted research blog than spend a month collecting filler links. Strong sources don't just pass conventional SEO value. They also shape how your brand is associated with topics across the wider web.

Create assets people actually want to reference

Most new sites launch with product pages, service pages, and a few blog posts. That's fine, but it doesn't usually attract meaningful links. If outreach is part of your growth plan, you need at least one asset designed to earn references.

Good launch-adjacent linkable assets include:

• Original research or benchmark writeups: Especially if you can speak from proprietary product or industry knowledge.
• Useful free tools: Calculators, templates, checkers, or generators.
• Definitive explainers: Pages that clarify confusing topics better than anyone else in the niche.

A practical example is a B2B ops platform launching with a migration checklist tool or a glossary that explains implementation terms in plain language. Those assets create reasons for industry blogs and consultants to reference the site.

Tie outreach to citation strategy

I like backlink campaigns that support answer-engine visibility, not just domain metrics. If a topic already triggers AI summaries, earning links to the page most likely to be cited is usually smarter than sending every outreach effort to the homepage.

The LLMrefs article on link building best practices is useful here because it aligns outreach thinking with broader AI-era visibility, not just old-school link acquisition habits.

A launch team can use that in a practical way. If the site has a strong “what is headless commerce” explainer and an okay feature page, outreach should support the explainer first if that's the page that answer engines are more likely to cite for category education.

9. Phase 3 Post-Launch A/B Testing for AI Performance

A lot of AI optimization advice is still opinion dressed up as certainty. The better approach is to test your own pages and build an internal playbook from observed patterns.

That doesn't require elaborate experimentation frameworks. It requires discipline. Pick one variable, create a variant, monitor how the page performs in both traditional search and AI citation contexts, then keep what improves clarity and visibility.

Test structure before you test style

For new sites, I'd start with page structure. It tends to have a bigger impact than copy polish. Compare a page that opens with a direct answer and a scannable step list against one that opens with brand narrative and abstract framing.

Useful variables to test include:

• Opening format: Direct answer first versus brand-led intro.
• Section design: Step-by-step guide versus general overview.
• Author visibility: Author credentials near the top versus lower on the page.
• Question coverage: Dedicated FAQ blocks versus no explicit Q&A.

A real example would be a legal-tech site testing two versions of a page about contract review software. Version A opens with a concise category definition and ideal use cases. Version B opens with product positioning. If AI systems keep citing the first version's structure, that pattern becomes part of your editorial standard.

Keep tests operationally realistic

Don't change everything at once. If the title, intro, schema, author block, and FAQ all change together, you won't know what mattered. That's why content testing works best when teams document variants carefully and review them on a steady cadence.

Testing should reduce opinion debt. If every launch decision still ends in “I think,” you don't have a system yet.

LLMrefs is helpful in this workflow because it gives teams a way to observe mention and citation changes around specific content themes. That makes post-launch optimization less subjective, especially when multiple stakeholders are arguing over what “better content” means.

10. Phase 3 Post-Launch Reporting and Continuous Improvement

A launch checklist is useful for one week. A reporting system is useful for the life of the site. Without one, teams repeat the same mistakes, misread short-term swings, and miss the patterns that drive growth.

I prefer reporting that combines technical health, organic performance, and AI visibility in one operating view. Not because every metric deserves equal weight, but because launch issues often overlap. A page can fail because it isn't indexed, because it's slow, because the intent is wrong, or because competitors are consistently more citable.

Build reports around decisions

The best launch reports aren't broad. They're answerable. Each report should help the team decide what to fix, what to scale, or what to stop doing.

A strong recurring view usually includes:

• Technical health: Indexation, crawl errors, blocked pages, and sitemap status.
• Organic movement: Landing pages, keyword groups, and conversions tied to search.
• AI visibility: Brand mentions, cited pages, and competitor presence by topic.
• Action log: What changed on the site since the last review.

For example, if a services page gets indexed quickly and starts ranking modestly but never appears in answer-engine citations, the next action may be content restructuring rather than link building. If another page is frequently cited but gets little click-through from search, metadata or positioning may need work.

Make reporting part of editorial operations

Here, many teams break the chain. They report to leadership, but they don't feed insights back into content briefs, design standards, or technical QA. Reporting becomes documentation instead of direction.

The stronger setup is monthly review with clear owners. The SEO lead handles crawl and indexation issues. The content strategist updates briefs based on citation patterns. The dev team addresses rendering or performance blockers. That's how a site improves after launch instead of just settling into whatever version went live.

10-Item New Site SEO Checklist Comparison

Item	Implementation complexity	Resource requirements	Expected outcomes	Ideal use cases	Key advantages
Phase 1: Pre-Launch Keyword & Content Strategy for AI	Medium, research and intent classification	SEO analyst, LLM tools (e.g., LLMrefs), 1–2 weeks	Conversational keyword map and prioritized long-tail questions	New sites preparing content for AI answer engines (e‑commerce, SaaS)	Aligns content to AI queries; uncovers long-tail opportunities
Phase 1: Pre-Launch Content Optimized for AI Citation	Medium–High, content + schema work	Content writers, schema tools, editorial review, 2–4 weeks	Content structured for direct answers and higher citation probability	Sites seeking immediate credibility and authoritative snippets	Increases chance of being cited; presents clear factual answers
Phase 2: Launch-Day Technical SEO & AI Crawlability	High, technical checks and rendering validation	Developers, SEO tools (crawlers, PageSpeed), 1–3 days	AI-accessible, fast-loading site with correct indexing	Launch readiness, JS-heavy sites, large catalogs	Prevents crawl/indexing issues; improves performance and crawlability
Phase 2: Launch-Day LLMs.txt Implementation	Low, policy file creation	Brand guidelines, LLMs.txt generator, <1 hour	Explicit instructions on citation, use, and attribution for models	Brands controlling attribution and training use of content	Quick to deploy; gives direct guidance to AI models
Phase 3: Post-Launch Brand Mention & SOV Monitoring	Low–Medium, setup and ongoing monitoring	Monitoring platform (LLMrefs), team access, ongoing	Share-of-Voice metrics and platform-specific visibility insights	Ongoing reputation tracking and competitive benchmarking	Real-time SOV tracking; highlights model-specific gaps
Phase 3: Post-Launch Content Gap Analysis	Medium, analysis and prioritization	AI citation tools, competitive data, 2–3 weeks initial	Identified high-opportunity topics, formats, and outreach targets	Filling gaps where competitors are cited by AI models	Targets proven AI-cited gaps; informs high-value content roadmap
Phase 3: Post-Launch Authority & E-E-A-T Signal Building	High, long-term credibility work	Experts, original research, PR/outreach, months	Stronger expertise/authority signals and trustworthiness	Sites addressing YMYL or seeking sustainable visibility	Builds durable authority; reduces misinformation risk
Phase 3: Post-Launch Backlink Strategy for AI Training Data	High, outreach and asset creation	PR, linkable assets (reports/tools), outreach, 6–12 months	High-authority backlinks likely to influence AI training signals	Brands aiming to appear in AI training corpora and top citations	Earns powerful trust signals from authoritative domains
Phase 3: Post-Launch A/B Testing for AI Performance	Medium–High, test design and analysis	A/B testing tools, analyst time, 4–6 weeks per cycle	Data on which headlines/structures get cited most	Optimizing content format, headlines, and author signals	Empirical optimization; builds repeatable content playbooks
Phase 3: Post-Launch Reporting & Continuous Improvement	Low–Medium, dashboards and processes	BI tools, API integrations, 1 week setup; ongoing maintenance	Regular insights, alerts, and prioritized improvement actions	Teams needing stakeholder reporting and iterative strategy	Continuous feedback loop; measurable impact tracking

From Checklist to Competitive Advantage

A new site often launches with clean design, approved copy, and a green light from stakeholders, then stalls because no one built for how discovery works now. The problem is not only rankings. It is whether the site can be found, parsed, trusted, and cited across Google Search, AI Overviews, ChatGPT, and Perplexity.

That is the difference between finishing a checklist and building an advantage.

The checklist matters because it forces the right work to happen in the right order. Before launch, teams need clear topic targeting, content that answers real questions, and page plans that avoid overlap. On launch day, they need pages that load well, render cleanly, and expose structure that search engines and answer engines can interpret. After launch, they need a system for checking what gets indexed, what gets cited, what gets ignored, and why.

That last part is where many launches fail. A site can have indexable pages and still disappear from AI-driven discovery if the content is vague, uncredited, hard to extract, or weaker than the sources answer engines already trust. I have seen technically sound launches underperform for months because the team treated AI citation as a future problem instead of a launch requirement.

Teams that outperform usually make a few disciplined choices early. They assign one job to each page. They write intros that answer the query instead of warming up for six sentences. They use headings, lists, tables, bylines, and source-backed claims where those formats help machines and humans interpret the page quickly. They also review which competitor pages get cited in AI results, then build something clearer, more current, or more defensible.

There are trade-offs. Publishing fewer pages with sharper intent usually beats launching a large library of thin content. Adding expert review slows production, but it strengthens trust signals on topics where credibility affects whether a page ranks or gets quoted. Strict template control can limit creative freedom, yet it often improves crawl efficiency, consistency, and extractability across the whole site.

Treat the first 30 to 60 days after launch as an observation window. Check server logs, Search Console coverage, crawl behavior, branded prompts in answer engines, and assisted conversions from organic landing pages. Rewrites are normal. Template fixes are normal. Cutting a page that should never have launched is normal too.

The advantage comes from closing that loop faster than competitors. Build for search rankings and answer engine citation from day one, then refine based on what the market rewards. That is how a new site stops being a project that shipped and starts becoming an asset that compounds.