Building the Marketing Engine Alongside the Product

Most B2B SaaS teams launch the product first and bolt on marketing six months later. The Semoria launch took the opposite route. The marketing engine was production-ready on day one: 47 blog posts in the index, three free tools live as lead magnets, persona landing pages across the priority niches, a content-calendar driven social automation system, and 27 transactional email templates wired into the trial and billing flows.

The decision was not aesthetic. It was operational. A SaaS without a marketing engine on day one wastes its launch window, and the launch window is the cheapest customer acquisition any product ever gets. This post walks through what shipped, why each piece earned a place in the six-week scope, and how a one-founder operation runs the entire engine without a marketing department.

What does Semoria's marketing engine actually include?

Six surfaces, each with its own role in the funnel:

Surface	Role in the funnel	What we built
Programmatic blog	Top of funnel, organic search	47 posts at launch, MDX with FAQ frontmatter rendered as JSON-LD FAQPage schema, direct-answer headings, AEO-ready structure.
Free tools as lead magnets	Top to middle of funnel	Voice Analyser, Post Generator, Voice Score Card. Three distinct value units, three distinct conversion funnels.
Persona landing pages	Middle of funnel, intent-aware capture	Programmatic SEO across priority niches with intent-aware copy and tier guidance.
Social automation	Top of funnel, distribution	Content-calendar YAML driving LinkedIn, X, and channel-specific repurposing.
Transactional email	Bottom of funnel and retention	27 templates covering trial onboarding, billing events, weekly digest, and feature discovery. Clickable stat tiles with UTM tracking.
Analytics and funnels	The loop that closes everything	PostHog funnels from first pageview through trial start to billing event. Conversion events on every meaningful step.

Each surface was scoped into the build from the start. The architecture decision record described the marketing engine in the same paragraph as the product. There was no separate "marketing phase" to be funded later, because the cost of building marketing surfaces inside the same codebase as the product is much lower than the cost of building them after launch on a separate stack.

How does the programmatic blog work?

The blog is a directory of MDX files in the codebase, each with a frontmatter block that carries the title, description, author, date, tags, category, hero image, and an faq array. The faq array is the most important part of the frontmatter because it serves two purposes at once: it renders as a visible "Frequently asked questions" section in the post, and it is also emitted as a JSON-LD FAQPage schema block in the page head. That second emission is the AEO-ready structure that Google and the AI search engines lift directly into their answer boxes.

The schema follows Google's published structured data guidelines for FAQPage exactly, which means the post is eligible for FAQ-style rich results in regular search and is a clean citation candidate for AI Overview. The visible FAQ section is a verbatim mirror of the frontmatter array, so the rendered page and the schema always agree.

The post body follows a consistent shape that holds across all 47 posts at launch. An opening hook that names the problem, an inline call-to-action that does not interrupt reading, four to six direct-answer H2 sections where the first sentence of each section completes the question in the H2, at least one HTML comparison table for the AI engines to extract, a mid-post banner call-to-action, a "Frequently asked questions" section, and a closing banner call-to-action. The structure is deliberate. It is the structure that ranks in Google's search fundamentals and it is also the structure that AI engines find easiest to summarise, which means the same post earns search traffic and citation traffic from the same investment.

Authority links, internal links, and an author byline with a Schema.org Person link to the founder profile round out the SEO surface. Backlinko's SaaS SEO research consistently lands on the same recommendations, and the Semoria blog was engineered against that body of work from day one.

How do free tools convert to paid?

Three free tools, each solving a contained slice of the problem the paid product solves, each exiting into the paid funnel cleanly:

The Voice Analyser at semoria.io/voice takes a sample of the user's existing LinkedIn writing, scores it against curated archetypes, and shows them the gaps. The output is a personal voice fingerprint they did not previously have, the five-minute payoff is real, and the natural next step is a 30-day free trial of the full Semoria product to act on what the fingerprint shows.

The Post Generator at semoria.io/tools/linkedin-post-generator drafts a single LinkedIn post against a topic and a target voice. It is rate-limited to five generations per day per IP to keep the cost-per-user negligible and to protect the trial signup as the natural upgrade path.

The Voice Score Card at semoria.io/tools/voice-score takes the Voice Analyser output one step further: it generates a 4:5 portrait card the user can share on their own LinkedIn, with their voice archetype and a seven-point radar. The card carries a discreet Semoria attribution and a UTM-tagged URL. Every share is a new top-of-funnel impression at zero distribution cost.

Each tool is a five-minute payoff that earns a 30-day free trial invitation without a hard pitch, and each one collects PostHog events on every meaningful action so the funnel is visible end-to-end. HubSpot's State of Marketing research consistently shows that high-utility lead magnets convert several times better than gated PDFs, and the Semoria tools were engineered against that finding. The conversion rate from free-tool engagement to trial start is the headline metric. Raw signup volume is not.

What are persona landing pages and how do they work?

The persona pages are the programmatic SEO surface. Each one lives at /for/[niche] and represents a specific intent: a job title, a niche, or a use case. The page combines a niche-specific headline, a relevance-led intro, the same product overview with copy tuned to the niche, and a CTA aimed at the right tier.

The data side is a single YAML file with one entry per niche carrying the slug, headline, intent description, tier recommendation, and copy fragments. The render side is a single page template that hydrates against the YAML entry. Adding a new niche is one block of YAML and ships with the next build.

This is the textbook shape of programmatic SEO done well: a structured intent dataset plus a template, not thin spun pages. Each page is genuinely useful for the niche it targets, has direct-answer headings, and links into the wider site. The long-tail traffic compounds over months without any further authoring work. The honest caveat is that programmatic SEO needs ongoing curation: niches that do not earn traffic after six months are removed or merged, and niches that earn meaningful traffic get a deeper investment.

How does the social automation work?

Social distribution is a content-calendar YAML file at the root of the marketing system. Each entry carries a date, a channel (LinkedIn, X, founder personal, Semoria company page), a content type (original post, blog repurpose, free-tool share), a draft body, and a status field. The schedule maps the cadence the GTM plan calls for, with blog repurposes seeded from the most recent published posts.

The repurposing layer is a set of channel playbooks that document the voice differences across channels. A blog post becomes a LinkedIn long-form, an X thread, a founder personal-voice anecdote, and a company-page educational post, each tuned for the channel. The same one-off authoring effort produces three or four distribution surfaces.

The calendar drives the work but does not autopublish. A human reviews and approves every queued post, the same human-in-the-loop discipline that runs across the product itself. Content Marketing Institute's annual benchmarks consistently land on consistency and channel-fit as the two factors that separate high-performing B2B content programmes from the rest, and a calendar-driven plus human-reviewed model is the cheapest way to get both at once.

How are transactional emails wired into the funnel?

Twenty-seven templates at launch, grouped into four flows:

Trial onboarding (seven templates): welcome, voice profile completed, first post drafted, first post published, midway nudge, trial-ending-soon, trial-ended. Each one carries clickable stat tiles linking back into the dashboard with UTM tracking.

Billing events (eight templates): subscription activated, plan changed, payment succeeded, payment failed, card expiring, cancellation requested, cancellation effective, win-back. Unambiguous on what happens next, because the cost of an ambiguous billing email is a support ticket or a refund.

Weekly digest (one template): Monday morning summary of posts published, scheduled, and pending review.

Feature discovery and announcement (eleven templates): in-app feature releases mirrored as targeted emails, plus admin-driven announcements. The bell-icon notifications in the app and the corresponding emails share a single data source.

Every template uses a shared component system (StatCard, FeatureShowcase, TipBox, AchievementCard, PlanCompare). The shared components are the difference between maintaining 27 templates and rewriting them. Gmail mobile compatibility quirks were nailed down once, in the shared components, and every template inherits the fix.

What does the analytics loop look like?

PostHog is the spine of the analytics loop. Every meaningful event on the site, the free tools, and the dashboard is tagged with a stable event name, the originating page, the tier the user is on, and the funnel stage. Funnels are defined for the four flows that matter: free-tool engagement to trial start, trial start to first post drafted, first post drafted to first post published, and first post published to billing event.

The funnel definitions are reviewed weekly, and the drop-off step on each one drives the next iteration of the relevant surface. The analytics loop is the part of the marketing engine that compounds over time, because each week's iteration is grounded in last week's data. A marketing engine without one is just a publishing programme.

How does this compare to other marketing setups?

The category-level comparison most B2B founders actually face:

Marketing setup	Surfaces live at launch	Ongoing cost	Time-to-first-trial-signup	Best for
Solo founder, no engine	Landing page, maybe a LinkedIn presence	Low	Months of distribution work to break out of zero	Pre-validation, hand-built early traffic
Typical agency-led setup	Brand site, a handful of blog posts, paid acquisition	£3,000 to £12,000 a month retainer	Faster, but expensive and slow to compound	Funded teams with budget but limited internal capacity
Semoria-style automated engine	Blog, free tools, persona pages, social automation, full email programme, analytics loop, all on day one	The hosting cost from the previous post, plus the founder's time	Days for the first traffic, weeks for compounding organic	Bootstrapped founders and small teams who want compounding marketing infrastructure on day one

The third row is what Semoria runs on, and it is the row that most independent UK SaaS founders should be defaulting to in 2026. The first row leaves the launch window on the table. The second row pays a retainer for surfaces that the founder could own outright on the same codebase as the product. The third row compounds because every blog post, every persona page, and every free-tool share keeps earning over time.

What does this prove about how Agenticise builds for clients?

The marketing engine on Semoria is the same engine Agenticise now uses for client platforms. The programmatic blog architecture, the FAQ-schema-driven AEO surface, the free-tool funnels, the persona pages, the content-calendar driven social automation, the shared-component email programme, and the PostHog analytics loop are all reproducible. The /case-studies/semoria hub is the long-form home for the full story. The origin post covers why Semoria exists at all. The methodology post covers how multi-agent orchestration in Claude Code compressed the timeline. The architecture post covers the lean modern stack the marketing engine runs on.

The agency now ships the platform and the engine that markets it, in the same six-to-eight-week window. That is the proof point. Marketing is not a phase that comes after launch. It is part of the same build, shipped in parallel, and ready on the day the product is live. If your team has been waiting on a marketing engine to follow a launch that has already happened, the same surfaces are still buildable, on the same stack, in a fraction of the time the conventional plan would take.

Frequently asked questions

Should I build marketing automation before launch?

Yes, when the automation is part of the lead magnet, the email follow-up, and the analytics loop rather than an afterthought. The mistake is to build the product, launch, then bolt on marketing six months later when the early traffic is already lost. The Semoria marketing engine shipped production-ready on day one because it was scoped into the build from the start: 47 blog posts at launch, three free tools as lead magnets, persona landing pages, social automation, and 27 transactional email templates wired into the trial and billing flows.

What is programmatic SEO?

Programmatic SEO is the practice of generating a large number of search-optimised pages from a template plus a structured dataset, rather than writing each page by hand. The dataset usually represents intent (a niche, a job title, a location, a use case) and the template renders it into a long-tail landing page with the right structure for search engines and the right copy for human readers. Done well, it captures search traffic across the long tail at marginal cost per page. Done badly, it produces thin content that the search engines correctly de-rank.

How do free tools convert to paid?

Free tools convert when each one solves a contained slice of the same problem the paid product solves, in seconds, with the user's own data, and exits cleanly into the paid funnel. Semoria's Voice Analyser scores a user's existing LinkedIn voice and shows them the gaps. The Post Generator drafts a single post in their voice. The Voice Score Card gives them a shareable archetype card. Each is a five-minute payoff that earns a 30-day free trial invitation without a hard pitch. The conversion rate from free-tool engagement to trial start is the metric to watch, not the raw signup volume.

How much can a small team automate in marketing?

Almost all of the repeatable work, and none of the editorial judgement. A single operator with the right stack can run a programmatic blog, three free tools, persona landing pages, social automation, and a full transactional email programme without ever owning a marketing department. The work that stays human is positioning, brand voice, the next big content bet, and any one-to-one conversation with a high-intent prospect. Semoria runs at exactly that ratio: one founder owns positioning and editorial; everything else is automated infrastructure.

What is the difference between AEO, GEO, and SEO?

SEO is search engine optimisation for traditional ranked results. AEO is answer engine optimisation, which means structuring content so it can be lifted directly into Google's AI Overview, Perplexity, or ChatGPT answers as a citation. GEO is generative engine optimisation, which is roughly the same idea applied across the broader set of generative search interfaces. All three reward the same underlying practices: direct-answer headings, FAQ schema, citation-grade sources, and content that is genuinely useful to the question being asked. The Semoria blog is engineered for all three at once.