saas.unbound is a podcast for and about founders who are working on scaling inspiring products that people love, brought to you by saas.group, a serial acquirer of B2B SaaS companies.
In episode #29 of season 5, Anna Nadeina talks with Josh, founder of Interact, a quiz builder specifically designed for lead generation.
Josh has been building businesses since 2008. He and his two UCLA co-founders launched Interact around 2011 with a simple-but-powerful idea: quizzes as a lead generation engine, not education. Growth was steady through 2018–2019, then stalled during the pandemic. The company stayed bootstrapped and focused on product-market fit, integrations and customer success — all of which set the stage for a rebound once AI became a factor.
How AI changed the go-to-market dynamic
Before large language models became mainstream, Interact’s growth primarily came from service providers — web designers and consultants who recommended Interact to their clients. The arrival of ChatGPT and similar tools initially disrupted that channel: business owners started turning to AI instead of agencies.
Josh explains the turning point: instead of fighting AI, Interact made AI a cross-selling partner. When users ask ChatGPT what to use for lead-generation quizzes, the model now recommends Interact and explains how to build a quiz in Interact’s platform. That shift helped Interact return to pre-pandemic growth trajectories.
AI inside the product (and what actually moves the needle)
- About 50% of Interact customers now use the built-in AI assistant to create quizzes — an agent inside the product that helps craft quiz content, suggest questions, and tune messaging.
- AI has improved onboarding, activation and retention. Interact is expanding from content agents to promotion agents and channel-specific assistants (website, social, etc.).
- But Josh is pragmatic: integrations (HubSpot, Klaviyo, and dozens more) are Interact’s primary differentiator. Product features come next, and AI is the third pillar. AI matters, but it’s not always unique — everyone can access the same models.
Where to place resources: integrations, product, AI
Interact’s allocation of engineering effort looks like this:
- Integrations engine (their own “Zapier” style connector set)
- Quiz builder and platform features
- AI (agents that assist with content, promotion, and implementation)
AI is integrated thoughtfully — where it interacts with the product (e.g., adding a question directly into a quiz) it provides real value. But for many tasks, general-purpose AI tools (ChatGPT, image generators) remain superior because they have broader capabilities like web search and multimodal memory.
How Interact gets recommended by ChatGPT and other LLMs
Josh’s approach is less about trying to “hack” the model and more about controlling the model’s source signals. Large models rely on widely available content sources: YouTube, blogs, review sites (G2), podcasts, social posts, and public conversations. To be the recommended answer, Interact focuses on being the most-cited, highest-quality voice across those channels.
- Off-domain presence matters: roughly 80% of Interact’s content strategy is off-platform (guest posts, reviews, social, podcasts); 20% is on their own domain.
- Off-platform advocacy is driven by customer success and proactive outreach: happy customers talking about the product is stronger than transactional link-building.
- Practical tactics include partnerships/trades (use my tool, I’ll use yours and we’ll both review), exceptional customer support, and asking satisfied users to leave reviews in places the AI is likely to pull from.
Rebuilding content for AI search: chunking for RAG
SEO for AI is fundamentally different from SEO for Google. Josh’s team revisited their top content and rebuilt it for retrieval-augmented generation (RAG):
- Write concise answer “chunks” — ~200 words each — that directly answer a specific question. Each chunk should be independently useful so a model can pull it into a response.
- Create long pillar posts by stacking many 200-word chunks. A post can be long (e.g., 10,000 words) but composed of many discrete Q&A-style blocks.
- Include data points and succinct, direct advice — the style a customer would expect from a real conversation.
In practice Josh’s team deleted and rewrote older posts rather than trying to translate them, because delivering direct answers is more effective for AI-based discovery than single-keyword pages.
Customer success, reviews and where to focus outreach
Interact over-indexes on customer success — two full-time support people in a team of nine — because word-of-mouth and public recommendations drive AI referrals. The playbook Josh recommends:
- Identify which customers are likely to talk about you on which platforms (e.g., Facebook groups, LinkedIn, niche communities).
- Prioritize making a small number of those customers “super happy” each month and ask them to write reviews or share their results.
- Use barters and mutual reviews with partners where appropriate — it’s efficient and relationship-driven.
“We’re nine people in an industry that has like $500 million of funding. Your reviews really help us show up where we otherwise can’t compete with advertising.”
Measuring the AI impact
Interact tracks AI-driven traffic and conversions and reports strong signals:
- ChatGPT became a meaningful referral channel starting July last year and has grown rapidly since.
- ChatGPT is now Interact’s fifth leading source of traffic and the third leading source of new trial signups.
- Referrals from ChatGPT have grown ~20–30% month-over-month for the past year.
Josh also notes that the true impact is larger than direct referral clicks: AI-driven discovery often leads users to search elsewhere (e.g., Google) and convert through other channels.
Product roadmap and AI features ahead
Key product directions Josh discusses:
- Get closer to the feel of a one-on-one customer consultancy inside the product — emulate multiple sessions, not a single linear chat.
- Ship multiple AI modalities (quick fixes, full quiz autogeneration, promotional planning assistants) so users can choose how much the AI does for them.
- Offer autogenerated quizzes that read a website and create a quiz instantly, while also keeping the assisted-build option for users who want more control.
Longer-term, Josh envisions AI assistants that can help iteratively across weeks — the equivalent of five hours of human coaching spread over short sessions, scaled with AI plus human support at key moments.
Wins, failures and human behavior around AI
Biggest win:
Scaling high-touch guidance using AI. Interact can now offer the depth of product coaching more cost-effectively, which drives better outcomes for customers and fuels growth.
Biggest failure / learning:
Building cool AI features without validating that customers actually want them. The lesson: understand human needs first — AI doesn’t change the need to deeply understand users.
Josh also observed that customer acceptance of AI is inconsistent — people are comfortable letting AI write headlines but sometimes balk at letting AI create core quiz questions. These preferences are often psychological and vary by user and task.
“Instead of hiring 10 people, you hire two people and eight AIs.” — a framing Josh found useful when organizing how to work with AI
Practical hacks and final advice
Josh’s favourite “hack” (learning borrowed from another founder): when you’re in a slump, talk to your customers. Hearing real success stories and how people use your product is a quick reset that informs product decisions and restores focus.
- Daily customer conversations = clarity on what to build, and whom to ask for reviews.
- Design product and AI experiences around real customer conversations, not hypothetical feature lists.
- Measure AI impact beyond direct clicks — consider how AI drives later-stage discovery and conversions.
Resources and where to learn more
Listen to the full episode for the full conversation and examples: saas.unbound Episode #29 with Josh Haynam.
Useful links mentioned in the episode:
- Interact: https://www.tryinteract.com/
- Josh Haynam (LinkedIn): https://www.linkedin.com/in/jhaynam/
- saas.group podcast channel: https://www.youtube.com/@saas-group
Table of Contents
Weekly newsletter
No spam. Just the latest news and articles from the world of SaaS and Acquisitions.