ARR vs. ERR: Why every dollar isn’t equal

Hello and welcome to The GTMnow Newsletter – the media brand of VC firm, GTMfund. Build, scale and invest with the best minds in tech.

Contracts have changed in the age of AI, and that means founders need to look at growth numbers through a new lens. Going from $0 to $2M in six months may sound impressive, but in practice that could be based on short-term pilots, with half of that revenue gone once the trials expire.

Instead of the 12-month commitments that defined SaaS for the last decade, many buyers are now asking for short pilots with easy opt-outs. These deals signal experimentation more than conviction, and as a result, a huge share of AI revenue today is what’s being called Experimental Runrate Revenue (ERR).

ERR is essentially unstable revenue from short pilots, opt-out contracts, or deals that only count once they convert. It can spike quickly as pilots stack up but just as easily collapse when those trials don’t turn into long-term commitments. That’s why contracting discipline matters more than ever. To turn experiments into durable revenue, you need to structure the deal the right way and then manage the process to ensure conversion.

This challenge is one that every AI company is facing, yet it’s rarely covered in the broader conversation. This edition breaks down both sides.

Harmonic helps you discover the best startups way ahead of the competition. We use it at GTMfund, as do thousands of investors at firms like USV and Insight. GTM teams at companies like Notion and Brex also rely on the platform to stay ahead.

Harmonic tracks millions of startups and lets you search using simple filters or natural language to match exactly what you’re looking for, so you can quickly evaluate and understand if it’s a fit.

At GTMfund, we even have a private Slack channel called #companywatchlist powered entirely by Harmonic.

Get 2 dedicated sessions with their team to help you test and structure the perfect searches here.

ARR vs. ERR: Why every dollar isn’t equal

New benchmarks are circulating, such as a16z’s latest growth data for early-stage companies, but topline revenue numbers don’t tell the full story. In the AI era, contracts look very different, which changes how we should think about Annual Recurring Revenue (ARR).

While you don’t need a refresher on ARR, it’s worth underlining why it matters: retention and recurrence. True ARR is predictable and easy to forecast. It becomes more profitable over time as customers – who were expensive to acquire – turn into stable, recurring cash flow.

This is where contracts in today’s AI landscape are impacting ARR. Instead of signing clean, 12-month commitments that defined SaaS for the last 15+ years, buyers are now asking for short pilots with easy opt-outs. The standard we’re seeing is a 3-month pilot with a pre-defined conversion to an annual contract. The customer can walk away at any time within the trial period with no penalty.

This new reality presents two key challenges:

1. How revenue gets reported. Many teams are booking pilot revenue as ARR in decks and updates, even when the customer isn’t paying full price yet and can churn in 12 weeks. It’s not surprising to see companies go from $0 to $1M ARR in 3 months with no visibility on how much is real ARR versus pilot dollars.

2. How customers see their commitment. From the buyer’s perspective, this isn’t ARR yet, it’s a trial balloon. They’re under pressure to test AI tools quickly, so they sign multiple pilots, run them across teams, and only keep the ones that stick.

This revenue is being framed as Experimental Runrate Revenue (ERR). ERR is inherently unstable – it can spike quickly as pilots stack up but can just as quickly collapse when those trials don’t convert.

The risk for founders is assuming ERR behaves like ARR. Pilots don’t guarantee long-term retention. Until they convert, treat ERR as a leading indicator of demand, not the durable foundation you can scale headcount and burn against. The market pull for some of these AI tools is unlike anything we’ve seen before, and that urgency is a huge advantage for every startup.

The go-forward strategy

The debate between ERR and ARR may not be as exciting as chasing historical growth rates, but it changes how deals are won, measured, and sustained. Revenue dynamics are shifting in ways founders and operators can’t afford to ignore.

1. Scrutinize your contracts

Every dollar isn’t equal. A $1.7M “run-rate” headline means little if half of it can evaporate next quarter. Go line-by-line and ask: What portion is true ARR versus ERR? Which contracts are fully committed, and which are still pilots with opt-outs? At Seed and Series A, investors are reading these carefully.

2. Pressure test the budget story

Ask customers (and yourself) where the budget is coming from. Is this spend locked in, or is it sitting in an “experimental AI” line item that may disappear? That’s not necessarily a bad thing (it’s often where adoption starts), but you need to know what milestones move you from “experimental” to “essential” in the buyer’s budget.

3. Align internally on revenue recognition

What do you call ARR? When can you book revenue, and how will you recognize it across the contract? Many first-time founders are navigating this for the first time. With more variability in pricing and contract structure, getting revenue operations right from day one is foundational.

The good news is there’s never been a more exciting time to build. But with growth and customer pull come quirks like evolving contract structures and experimental budgets. As those experimental budgets harden into “must-have” line items, founders who treat revenue recognition as a strategic discipline will be in the best position to scale sustainably.

Share

Tag @GTMnow so we can see your takeaways and help amplify them.

GTM 161: How to Build a GTM Roadmap

Paul Williamson (ex-Plaid, $3M → $300M ARR) on sequencing GTM bets, spotting high-value clients, and building forward-compatible roadmaps.

Listen on Apple, Spotify, YouTube or wherever you get your podcasts by searching “The GTMnow Podcast.”

Coffee just made Salesforce and your GTM stack chat-native. With their new MCP interface, you can now use ChatGPT 5 or Claude to pull answers directly from Salesforce, HubSpot, Google Workspace emails and meetings, and any enrichment data tied to your contacts and accounts. Instead of clicking through dashboards, you can just ask questions like, “Which deals in my pipeline are stuck past 30 days?” or “Show me all enterprise accounts with open support tickets.” The answers come back instantly, right inside your AI assistant.

AngelList’s H1 2025 report shows venture slowly stabilizing after a turbulent few years. AI and Robotics continue to dominate, drawing nearly 60% of all capital. Seed rounds are holding steady at around $20M, Series B has climbed to $400M, and while many COVID-era startups are still under pressure, the 2023–24 AI cohorts are emerging as the clear winners.

1. Manager, Sales Development – EMEA at Vanta (Hybrid – Dublin, Ireland)

2. Director of Customer Success at Vividly (Remote – USA)

3. Growth Marketing Manager at Alt (Remote – USA)

4. Enterprise Customer Success Manager at Gorgias (Remote – Vancouver)

5. Vice President, Sales at Amper (Remote – USA)

See more top GTM jobs on the GTMfund Job Board.

If you’re looking to scale your sales and marketing teams with top talent, we couldn’t recommend our partner Pursuit more. We work closely together to be able to provide the top go-to-market talent for companies on a non-retainer basis.

Upcoming events you won’t want to miss:

• GTMfund Dinner (private registration): September 10, 2025 (San Francisco, CA)

• Pavilion GTM Summit: September 23-25, 2025 (Washington, DC)

• National Mall Walk with GTMfund: September 25, 2025 (Washington, DC)

• Dreamforce: October 14-16, 2025 (San Francisco, CA)

• GTMfund Dinner (private registration): October 22, 2025 (Austin, TX)

• GTMfund Dinner (private registration): November 18, 2025 (Toronto, ON)

• GTMfund Dinner (private registration): November 19, 2025 (New York, NY)

• GTM x Founder Event (private registration): November 20, 2025 (New York, NY) – if you’re an AI-focused founder in NYC, hit reply to get the details

Some GTMnow community (founder, operator, investor) love to close it out – we appreciate you.

Subscribe now

This newsletter was written and edited by Sophie Buonassisi, Tetiana Paratsii, Paul Irving, Max Altschuler and the GTMfund team (not AI!).