Claimlane Claimlane Claimlane Claimlane

Let prospects see Claimlane with their own products

It is one thing to show someone a generic product demo. It is much easier to understand when it looks like your own business.

So I built a tool where you can enter a store URL and get a working Claimlane portal using that company's logo and products.

There are two versions. One lives on the website and captures leads in HubSpot. The other is used by sales, so they can create a personalised version and send it directly to a prospect.

We also send activity to Slack, which means the sales team can see when someone has actually opened the demo.

• Cloudflare Workers
• KV
• Vanilla JS
• HubSpot
• Slack

A pricing-page ROI calculator that pays back its build cost

Four sliders, one email gate, a real number at the end. Prospects on the pricing page get a personalised savings estimate — and we get a lead qualified before the SDR ever picks up the phone.

Submissions land in HubSpot with hidden lead-source fields, a personalised PDF report goes out on submit, and the ROI signal was wired into our lead scoring within a week of launch.

• Vanilla JS
• HubSpot
• MQL scoring
• Slack

Reworking 160+ pages and changing how we approached SEO

We had more than 160 pages that needed attention, so I worked through them one by one.

The goal was not just to rewrite the copy. We introduced clearer rules around how pages should be built: use real customer examples, verify internal links, remove unsupported claims and avoid the language that makes B2B content all sound the same.

We also reduced our publishing pace from around 20 articles a week to roughly six.

Publishing more was not necessarily helping. In several cases, we were creating pages that competed with each other for the same searches.

Alongside the content work, I worked through technical SEO issues including broken links, redirects, metadata and schema, and briefed the external SEO agency we work with.

• Semrush
• Webflow
• Claude
• Notion

Shipping 350+ articles in a year without losing our voice

We started at ~5 blog posts a month written by hand. That didn't scale to the SEO ambition we had.

So I built the pipeline: Semrush pulls 6,000+ target keywords → a Notion writing agent drafts each one using our tone of voice and internal-link rules → drafts land in Webflow with FAQ schema and cluster tagging → I QA and publish.

At peak it ran 10 articles a week; May 2026 was a 77-article month. Around 355 posts landed during the year, and 18 of Claimlane's top-20 traffic keywords now live on pages this system produced.

• Semrush
• Semrush MCP
• Notion
• Claude
• Webflow
• n8n

Getting cited inside Google AI Overviews, on purpose

In October 2025 I added an llms.txt to the site and started the schema markup work. Most B2B marketers hadn't heard of GEO yet; I wanted us in the answer set when buyers stopped clicking blue links.

In May 2026 I put together the AI Visibility Action Map: 45 real buyer prompts, 22 topic clusters, and a rewrite pass on the top money pages to match how AI answer engines assemble responses.

Result: Claimlane is now cited inside Google AI Overviews on 10 of its top-20 US keywords, including "warranty management software", "rma number", "delivery exception" and "ecommerce returns".

• llms.txt
• JSON-LD schema
• Claude
• Manual prompt research

Turning sales calls into useful feedback automatically

We record a lot of sales calls, but there is only so much time managers can spend listening back to them.

I built a workflow that takes each recorded call, reviews it against the sales framework we use and sends a short summary to Slack. The relevant notes are also added back to HubSpot.

That gives reps feedback while the conversation is still fresh instead of weeks or months later.

I've built several smaller workflows around the same idea, including meeting preparation, lead screening, enrichment and weekly growth reporting.

I've found that the simplest automations are often the ones that end up being the most useful.

• n8n
• Fireflies
• Anthropic API
• HubSpot
• Slack

What we learned from more than 600 sales calls

We had years of recorded sales calls, but we had never really looked at them together.

So I went through more than 600 conversations and pulled out the moments where prospects talked about their problems, frustrations and reasons for looking for something new.

A few themes kept coming up.

People rarely said, "We need returns software."

They talked about not knowing what was happening with a case. Spending too much time on manual work. Moving information between systems. Sending too many emails. Or losing customers because a warranty claim had been handled badly.

Those findings helped shape both our website messaging and the companies our SDR team focused on.

• Python
• Fireflies
• Claude

One of the first B2B teams to run ChatGPT ads

OpenAI opened Ads Manager for advertisers in June 2026. I had Claimlane's account set up on day one and started designing tests immediately.

July was the first proper run: 6 campaigns, email-account targeting, full conversion tracking wired to our attribution stack. August was a conversion-quality pass — figuring out which of the leads that clicked actually behaved like real prospects vs. curiosity clicks.

Most B2B marketers are sitting on the sidelines waiting for case studies. Being early gives us weeks of learning on a channel that's about to get very expensive.

• OpenAI Ads Manager
• GA4
• HubSpot conversion tracking

Making it easier for everyone to create on-brand material

Our decks and one-pagers were starting to look slightly different depending on who made them.

Different blues. Different type sizes. Old screenshots. Old logos.

So I helped set up a simple design system in Figma with shared colours, typography and reusable assets.

I also created templates that sales can duplicate and edit themselves instead of needing someone from marketing every time they want to change a slide.

A lot of the repetitive Figma work was done through the API rather than manually.

• Figma
• MCP
• Figma Plugin API

Helping the team actually use AI after the first workshop

Getting people excited about AI for an afternoon is easy.

Getting them to keep using it a month later is harder.

After our first internal session, everyone had built something. The next challenge was making sure those ideas actually got used.

So I set up a simple system around it: a Slack channel for sharing, a board where everyone tracks what they are building, a backlog of ideas and a weekly session where someone shows what they made.

The most important part is sharing.

If one person builds something useful, the next person should be able to use it rather than starting again from scratch.

• Notion
• Slack
• Claude