Inside SunHouse’s 2026 Marketing Tech Stack: AI, Analytics and Execution

Turning the Analysis Into a Usable Artifact

Another important part of my workflow is turning the final analysis into a usable artifact.

That artifact might be a structured report, visual brief, wireframe, annotated recommendation, implementation document, presentation, or another format designed around the person who needs to take the work forward.

For example, I recently completed a CRO review of a client’s website. I am not the designer responsible for rebuilding the pages, so the useful endpoint was not a long conversation explaining what I thought should change. I turned the findings into artifacts the designer could work from, with the recommendations organized and translated into a clearer implementation brief.

I apply the same principle elsewhere. Strategy should become something a team can execute, research should become something a client can understand, and analysis should be translated into a format that a designer, developer, media buyer, copywriter, or project manager can actually use.

For me, that is an important distinction in using AI well: the final output should be designed for the person who needs to act on it, not for the AI conversation that produced it.

The principle behind all of this is simple: context before prompting. The better the model understands the business, the evidence, the constraints, and the process, the stronger and more useful the output becomes.

ChatGPT

DAILY

Writing · Analysis · Research · Critical Review · Problem-Solving

I pay for a ChatGPT Plus subscription. ChatGPT is another daily working environment for me, and it currently supports most of my writing projects. While Claude has become the broader operating environment for much of my AI work, ChatGPT has become particularly important for writing because I have spent a great deal of time refining the overall voice.

Over time, I have been very deliberate about teaching it what good writing sounds like to me: natural rather than formulaic, authoritative without becoming stiff, conversational where appropriate, and free of many of the patterns that make AI-generated copy immediately recognizable. I have also defined the language, structures, phrasing, and habits I want it to avoid.

For many writing assignments, ChatGPT now gives me a stronger starting point because the underlying voice is already much closer to what I want. I still edit and challenge the work, but I spend less time stripping away generic AI language before I can get to the substance.

Projects and Context

As with Claude, I do not use ChatGPT as one undifferentiated stream of conversations. Any ongoing client, initiative, research stream, or substantial body of work gets its own Project.

Each Project has its own instructions and relevant background material so the model does not have to reconstruct the business from individual prompts. Those instructions typically establish the client’s positioning and audience, brand voice, strategic assumptions, marketing terminology, writing standards, constraints, evidence requirements, preferred structure, and anything I specifically want challenged rather than simply accepted.

The broader writing voice provides consistency across my work, while the Project instructions add the context that is specific to the client. That allows a healthcare organization, B2B company, ecommerce business, nonprofit, and SunHouse itself to sound distinct without starting from scratch every time.

I also provide source documents, previous work, research, client materials, and other relevant context when they improve the assignment. As with Claude, I would rather give the model the information it needs than allow it to fill important gaps by inference.

Writing and Iteration

I use ChatGPT throughout the writing process, not simply to generate a first draft.

Sometimes that means helping me organize an argument before I write. Other times I may bring in a rough draft, dictated thoughts, research, client feedback, or an existing piece and work through it iteratively.

A large part of the value is the back-and-forth. I can challenge a sentence, explain why something does not sound like me, ask for a stronger marketing argument, question whether a paragraph is strategically sound, or keep refining until the language and thinking are both where they need to be.

My instructions, examples and retained context help keep subsequent drafts closer to the voice I want. I do not simply accept or reject an output. I explain what is wrong with it and what I would do differently, which makes the working environment more useful over time.

Critical Review and Second Opinions

ChatGPT also plays an important role as a critical second opinion.

I regularly bring it strategies, drafts, recommendations, research findings, and analyses developed elsewhere and ask it to pressure-test the thinking. Rather than asking a series of disconnected questions, I want it to examine the reasoning as a whole: identify weak assumptions, find gaps, consider credible counterarguments, question whether the evidence actually supports the conclusion, and point out what an experienced specialist might challenge.

I use Claude in the same way, often taking work from one environment into the other.

Agreement between two models is not independent verification. For important decisions, I check the evidence rather than treating the first AI response as the answer. I compare perspectives, introduce opposing arguments, add evidence, and keep refining the reasoning until I am comfortable using it to support an actual marketing recommendation.

That is an important part of how I use AI generally. The value is not simply producing content faster. It is having another sophisticated layer of writing, analysis, and critique available throughout the work.

Perplexity

REGULAR ROTATION

Sourced Research · Source Discovery · Verification

Perplexity is my first choice when the quality of the research and the sources behind it matter.

In my own workflow, it consistently does a better job of finding strong, relevant source material and giving me a clear trail of citations to work from. That makes it particularly useful when I am researching a market, validating a claim, looking for original studies or data, investigating competitors, or building the research foundation for a strategy or article.

I tend to use ChatGPT and Claude differently. They are often stronger environments for thinking through the research, challenging it, synthesizing multiple sources, and ultimately turning it into strategy or writing. But when my immediate question is “Where is the strongest evidence for this?”, I usually start with Perplexity.

That does not mean I accept its citations at face value. An AI-generated answer is not the source, and the presence of a citation does not guarantee that the source actually supports the claim being made. I still open the underlying material, check how current and authoritative it is, look for the original source where possible, and distinguish primary research from secondary articles repeating someone else’s findings.

For research-heavy work, that source trail is one of the main reasons Perplexity continues to earn a place in my stack.

Fyxer

DAILY
Email Triage · Inbox Management

Fyxer helps manage part of the administrative layer around email.

The value is less about email itself and more about protecting senior time from routine inbox management so more attention can go toward strategy, clients, and decision-making.

Zapier

DAILY
Workflow Automation · Lead Routing · Data Movement

Zapier is one of the connective pieces of the SunHouse stack.

We use it when two systems need to communicate but the workflow is not available natively.

That might involve:

• Passing a lead into a CRM
• Triggering an email sequence
• Creating a task
• Sending an internal notification
• Moving campaign data
• Updating customer information
• Connecting a form with another application
• Routing information based on specific conditions

The best automations are often the least glamorous ones: repetitive manual work that should never have required a person in the first place.

But automation also needs safeguards.

We think about duplicates, incomplete data, failed steps, attribution fields, conditional logic, and what happens when the expected workflow breaks.

Automating a bad process simply creates mistakes more efficiently.

SEO, GEO, Local SEO, and AEO

Semrush

DAILY
SEO · Competitive Intelligence · GEO/AEO · Content Strategy

Semrush is one of the most heavily used tools in our search stack.

I use it for:

• Keyword research
• Competitive analysis
• Backlinks
• Rankings
• Content gaps
• Technical SEO
• Search visibility
• Competitor traffic analysis
• Content opportunities
• AI visibility research

I make an important distinction between Semrush data and first-party data.  Semrush gives us large-scale modeled data about a market, keywords, competitors, rankings, backlinks, and traffic.  Google Search Console and GA4 report on the client’s own digital properties, within the limits of their tracking, attribution and reporting coverage.

Both are valuable, but they answer different questions.

I also connect Semrush directly into Claude through its MCP integration where appropriate. That changes the workflow considerably.  We can increasingly bring current marketing data directly into the environment where the analysis is happening.

For GEO and AEO, I use Semrush and other research sources to examine questions such as:

• Which brands are appearing in AI-generated answers?
• Which competitors are being mentioned instead of the client?
• Which sources are being cited?
• Which pages are earning citations?
• Which questions is the brand failing to answer?
• Where is authoritative content missing?
• Which third-party sources influence the answer ecosystem?
• Is the brand’s entity information clear and consistent?
• Which existing pages should be expanded or restructured?

That research informs our SEO and AI visibility work, including content priorities, internal architecture and authority building.

One practical implication is that I am increasingly focused on whether search engines and AI systems can clearly understand who a brand is, what it is authoritative about, and where that authority is corroborated. Entity clarity, topical depth, internal architecture, structured data, and credible third-party mentions all contribute to that picture. I am much less interested in supposed GEO hacks that sit outside the fundamentals. Google’s current guidance explicitly says that websites do not need special AI text files or AI-specific markup to appear in Google’s generative search features.

There are also practitioner case studies worth examining. Search Engine Land documented its own 2026 case study showing gains in organic visibility and AI Overview citations after strengthening topical authority and Knowledge Graph/entity optimization. The author co-created TopicalBoost, the product evaluated, so I treat this as a disclosed practitioner case study rather than independent validation.

Google Business Profile

REGULAR ROTATION
Local SEO · Maps Visibility · Reviews · Local Presence

For local clients, we manage Google Business Profile directly. That includes keeping core business information accurate, reviewing categories and services, managing photos and updates, monitoring reviews, and making sure the profile supports the broader local SEO strategy.

Semrush remains our primary local SEO research and monitoring tool, while Google Business Profile is where much of the actual local presence is managed.

BrightLocal

SPECIALIST
Local SEO · Reputation · Local Search Visibility

I have used BrightLocal occasionally for clients where local reputation and local-search performance are especially important. It provides useful local rank tracking, citation, reputation, and Google Business Profile-related capabilities, although it remains a specialist option rather than a core part of our stack.

Paid Media and Creative Testing

Google Ads, Meta Ads, LinkedIn Ads and Microsoft Advertising

These platforms support different audiences and buying situations. We choose channels around the offer, the customer and the economics of acquiring a qualified lead. Campaign decisions need conversion data and sales feedback, as well as clicks and platform-reported results.

Ryze AI adds an optimization layer, while AdCreative.ai supports creative production and testing. I evaluate recommendations against the account context and business objective. Our paid media work includes the offer, landing page and measurement needed to judge whether a campaign is producing useful demand.

I have also been testing ChatGPT Ads for SunHouse. It is an experiment in a developing discovery channel, with results assessed separately from established client acquisition channels.

Analytics, Attribution and Reporting

GA4, Search Console, Tag Manager, CallRail and Custom Dashboards

GA4 helps us examine acquisition and conversion behavior. Search Console shows search queries, page performance and indexing information. Tag Manager supports event implementation, and CallRail adds call attribution. I am planning to phase out Data Studio (formerly Looker Studio) and build my own dashboards. That transition is planned rather than complete.

A form submission, a qualified inquiry and a customer are different outcomes. CRM records and sales feedback help connect marketing reporting to what happens after the initial conversion.

Landing Pages and Conversion Optimization

Unbounce, Contentsquare (formerly Hotjar) and VWO / Wingify

Unbounce supports landing page production and iteration. Contentsquare (formerly Hotjar) helps investigate visitor behavior, while VWO / Wingify supports structured experimentation where the traffic and test conditions justify it. We use the combination appropriate to the client and question.

A recent website CRO review illustrates how this work fits together. I developed recommendations on the consultation offer, copy and form placement, then organized those findings into a brief for the designer. Our conversion rate optimization work needs to leave the person implementing it with clear priorities and actionable recommendations.

CRM, Email and Lifecycle Marketing

Salesforce, Zoho and Email Platforms

Salesforce and Zoho CRM supply customer and sales context. Email and automation work may use Zoho Campaigns, Zoho Marketing Automation, ActiveCampaign or Klaviyo, depending on the client’s existing systems and requirements. Zoho Campaigns and Zoho Marketing Automation are separate products.

One practical application is separating contacts who have never visited, visited without purchasing, and already purchased. Each group needs different follow-up. Another is distinguishing past customers, older inquiries and opportunities that did not close. The work involves segmentation, copy, timing, CRM fields and checking whether the sequence produces a response.

NeverBounce supports list hygiene. Zapier connects steps where native integrations do not cover the process. We still need to check consent, deliverability, duplicate records and failed automations.

Creative Production and Collaboration

Canva, Midjourney, Metricool and Google Workspace

Canva supports creative production, Midjourney comes in for selected image-generation work, and Metricool supports social scheduling and management. Google Workspace remains part of day-to-day documentation and collaboration. AI-generated visuals still need review for brand fit and suitability.

How the Stack Supports Marketing Leadership and Execution

In my fractional CMO work, I use these systems to decide what deserves attention, what evidence is missing and what the team should do next. That includes setting priorities, defining measurement and reviewing execution with the people responsible for delivery.

Through SunHouse’s agency services, our team carries out the campaigns, content, tracking, conversion improvements and lifecycle marketing those decisions require. The tools help connect the work; the recommendation still needs an owner, a reason and a way to evaluate the result.

If you are considering the right level of support, our guide to fractional CMO service models explains the available approaches. You can also speak with SunHouse about marketing leadership or implementation.