Why did Apollo's sales-led model stop working?
Its price sat in the wrong place: too expensive for the small businesses that wanted Apollo, and too cheap to pay for the sales team selling it. That is how Tim Zheng, Apollo.io's co-founder and CEO, explained it in a SaaStr talk published on 12 September 2024. Figures below come from that talk unless marked otherwise.
Apollo started out entirely sales-led. SDRs generated pipeline, account executives closed deals, and customer success managers onboarded and expanded accounts. The first years went well, Zheng said: annual recurring revenue grew from $0 to $2 million, then to $5 million. Then growth stopped, right around 2019.
- Cash. Apollo had raised $9 million in its Series A and was down to its last million, with under six months of runway.
- People. It had 50 employees, morale was low and good people were leaving.
- Competition. Zheng said rivals in a crowded sales tech market shipped far faster than Apollo could.
- Unit economics. Because SDRs, account executives and customer success managers are expensive, Apollo was spending about a dollar to acquire 80 cents of revenue.
Zheng's diagnosis went past product-market fit to what Zheng called product-market-model-channel fit. Apollo had product-market fit, because small and mid-sized businesses (SMBs) liked its all-in-one product. It lacked model fit, because contracts of about $10,000 a year were too expensive for most SMBs. It lacked channel fit too, because $10,000 was not enough to pay for a sales-driven motion. Zheng called that price band a dead zone.
What did Apollo change, and when?
Apollo cut its contract value by about 90%, made the product fully self-serve with a free plan, and shrank to fit. On timing, Zheng's talk places the stall around 2019 and describes a two-year transformation, without giving a launch date for the free plan. Bain Capital Ventures, which led Apollo's 2023 Series D, wrote in August 2023 that Apollo first required custom work for every engagement, and that a major product overhaul made it completely self-serve, with a free tier.
| Before (sales-led) | After (product-led) | |
|---|---|---|
| Price | About $10,000 a year per customer | About $1,000 a year per customer; free plan at the entry |
| How customers buy | SDR, account executive, then customer success | Sign up, use, then pay; sales talks mostly to people already using the product |
| Top of funnel | Hundreds of leads a month | Tens of thousands of sign-ups a month needed |
| Team | 50 people | 10 people; Zheng asked the co-founders whether they would go to zero salary until Apollo grew again |
| Versus rivals | Sold through a sales team, like most rivals | The opposite of rivals, who mostly sold point solutions to mid-market and enterprise buyers at about $30,000 a year |
Pavilion's notes for Zheng's June 2025 Topline podcast episode summarise the same move as cutting prices to $99 a month and going all in on product-led growth. Zheng gave the reason for cutting the team in the talk: getting to cash-flow positive was the only way to survive and keep a core group committed to a multi-year change.
By 2023 Zheng was pitching the model against rivals. Zheng told TechCrunch in August 2023 that Apollo "is free to get started and, with our product-led growth motion, allows monthly pricing rather than expensive multi-year contracts."
How did Apollo test a 90% price cut?
With an A/B test on new traffic only. Cutting the price by 90% would hit every existing customer at renewal, so Zheng said the team first tested the new pricing on new visitors. Those who saw the new pricing converted at 20 times the rate, even though the price was 10 times lower.
That result settled the decision, and it was not free. Zheng described the following year's renewal season as brutal, with customers downgrading by about 90%. Zheng did not give the test's sample size or dates; these figures are Zheng's, from the talk.
The rest of the turnaround came from a modelled target and a repeatable method. Zheng said the team set a goal to grow revenue 10 times, to $50 million, and worked backwards: retention had to rise 5x, monetisation 10x and sign-ups 50x. The method, which Zheng calls a scientific method for moving metrics, has five steps:
- Break the metric you want to move into sub-metrics.
- Pick the sub-metric with the most room to improve.
- Brainstorm and prioritise ideas.
- Ship, then measure.
- Repeat until the metric moves, then go to the next one.
Zheng said at most about half the experiments worked, and that the gains came mostly from improving existing features, with few new ones. Using the same method on feature gates, plus an A/B testing framework for pricing variants, Zheng said, conversion to paid improved 10x over a year.
How did self-serve Apollo find users without an ad budget?
Through search pages built from its own data and through word of mouth. Zheng said paid ads did not work at Apollo's lower prices, and the team could not afford content writers.
- A page per contact. Sales reps search for information about the people they want to reach, and Apollo had hundreds of millions of contacts in its database. One engineer spent four weeks building a public page for each, indexed on Google. Zheng said it brought orders of magnitude more traffic than paid ads.
- Net Promoter Score as a growth lever. Referral credits were not enough on their own, Zheng said, so the team worked on NPS instead, fixing the reasons people scored low. NPS rose from 5 to 60 over about two years, and by the time of the talk over 90% of sign-ups came from organic word of mouth.
- Treat human help as a bug. Borrowing a principle Zheng credited to Atlassian, the team listed every reason customers needed a sales, support or success rep, then removed the low-value ones with product changes or marketing videos. Zheng said both acquisition cost and conversion improved.
Bain Capital Ventures quotes Zheng's rule for this in August 2023: customers "can get value out of it without needing to talk to anybody."
What did the switch produce?
Rapid growth, by the figures Apollo and the press reported at each step:
| Date | Figure | Source |
|---|---|---|
| March 2022 | Over a million users across 160,000 companies signed up in the previous year; paying customers up 60% in one quarter, from 9,000 to over 15,000; $110M Series C led by Sequoia Capital | Apollo's Series C post (company-reported) |
| March 2022 | $900 million post-money valuation; paying segment of 16,000+ businesses | TechCrunch |
| August 2023 | Revenue up 900% since 2021; three million users across 500,000 customers, 40,000 of them paying; $100M Series D at a $1.6B valuation, led by Bain Capital Ventures | TechCrunch |
| September 2024 | Revenue up over 25x since the change began, by Zheng's account; most of the work done by a team of 10 to 15 | Zheng's SaaStr talk |
| June 2025 | $150M+ ARR | Pavilion's Topline episode notes |
The free entry point is still there. Apollo's pricing page, checked on 20 October 2026, lists a Free plan and says its free plan stays free with no time limit.
The lesson Zheng drew for small teams in the talk: a team of 10, with no more than 15 people, did almost all of this by being focused and moving fast.
How can a small tool test a no-sales-call entry plan?
Our advice is to run a smaller version of Apollo's test: offer a self-serve plan to new visitors only, keep the sales path for everyone else, and compare revenue per visitor after 30 days. First, check whether your price can carry a sales call at all. Varun Anand, co-founder of Clay, told First Round Review in January 2025 that a sales-led motion needs at least a $5,000 to $10,000 annual contract, and Zheng put Apollo's $10,000 contracts in the dead zone. If most of your deals are well below that and every one needs a call, the call may cost more than the deal is worth.
Write the entry plan
One plan, monthly, card payment, a clear usage limit, and nothing that needs a call to start. Keep a booking link for teams that want one.
Split new traffic only
Show half of new visitors the current path (book a call) and half the self-serve plan. Leave existing customers on their plans during the test; Apollo's renewal season shows why.
Set the stopping rule in advance
We suggest four weeks or 100 visitors per arm, whichever comes later. With small numbers, two customers against four tells you nothing.
Log every reason someone asks for a call
Tag each one high value (a strategic question) or low value (a how-to question). Fix the top three low-value reasons in the product or with a short video, as Apollo did.
Decide on revenue per visitor
Compare 30-day revenue per visitor and time spent per customer. Conversion alone hides a lower price.
| Measure | Arm A: book a call | Arm B: self-serve plan |
|---|---|---|
| Visitors | Count | Count |
| Paying customers after 30 days | Count | Count |
| Revenue per visitor | 30-day revenue divided by visitors | 30-day revenue divided by visitors |
| Your hours per new customer | Calls, follow-ups, onboarding | Support questions answered |
For arm A, a booking page that reads your real calendar keeps the call path quick; Siela's booking pages check every connected calendar and send reminders. If arm A wins, tighten who gets a call with our guide to qualifying a lead fast.
Frequently asked questions
Why did Apollo switch from sales-led to product-led growth?
Tim Zheng said in a September 2024 SaaStr talk that Apollo's contracts of about $10,000 a year were too expensive for the small businesses that liked the product and too cheap to pay for SDRs, account executives and success managers. Apollo was spending about a dollar to acquire 80 cents of revenue and had under six months of runway.
When did Apollo launch its free plan?
Apollo has not given an exact date in the sources we checked. Tim Zheng's September 2024 talk places the stall around 2019 and describes a two-year change to self-serve. By March 2022, TechCrunch reported that Apollo operated on a freemium model, so a free plan existed by then.
How much did Apollo cut its prices?
About 90%. Tim Zheng said Apollo moved from contracts of about $10,000 a year to about $1,000, after an A/B test on new visitors showed the lower price converting at 20 times the rate. Existing customers downgraded by about 90% at renewal. Pavilion's notes for a June 2025 podcast episode say Apollo reduced prices to $99 a month.
How did Apollo get sign-ups without paid ads?
Tim Zheng said one engineer spent four weeks building a public, indexed page for each contact in Apollo's database, which brought far more traffic than paid ads. The team also raised its Net Promoter Score from 5 to 60 over about two years, and by September 2024 over 90% of sign-ups came from organic word of mouth.
What is product-market-model-channel fit?
It is the four-part test Tim Zheng used to explain Apollo's stall. Product-market fit means the market wants the product. Model fit means the pricing suits that market. Channel fit means the way you sell is affordable at that price. Apollo had the first, but a $10,000 contract fitted neither small buyers nor a sales team.
Sources
- SaaStr AI (YouTube): Transitioning from Sales-Led to Product-Led Growth and Scaling to $100M ARR with Apollo's CEO (published 12 Sep 2024)
- TechCrunch: Apollo.io, a full-stack sales tech platform, bags $100M at a $1.6B valuation (published 29 Aug 2023)
- TechCrunch: Apollo.io raises $110M as its sales intelligence platform crosses 16K paying companies (published 3 Mar 2022)
- Apollo: Apollo.io Secures $110M Series C Funding Led by Sequoia Capital (Tim Zheng) (published 3 Mar 2022)
- Bain Capital Ventures: How Apollo.io Is Helping Every Company Reach Its Full Revenue Potential (published 29 Aug 2023)
- Pavilion: How Apollo Plans to Grow From $150M to $1B+ in ARR (Topline episode notes) (published 24 Jun 2025)
- Apollo pricing (checked 20 Oct 2026)
- First Round Review: The GTM Inflection Points That Powered Clay to a $1B+ Valuation (published 23 Jan 2025)