What is revenue operations at a hypergrowth company? Doug Hanna of Grafana Labs, part 2

If you want a concrete answer to what is revenue operations, this is the episode. Doug Hanna returns to walk through the commercial machine at Grafana Labs: how the sales organization is cut, what the customer journey actually looks like, and what changed at each stage of growth.

Part one covered scaling the organization from 70 people to more than 500. This is the other half, the go to market side, and it is considerably less tidy than the outcome suggests.

What he walked into

Hanna joined in September 2019, just barely pre Series A. The company was close to term sheets and investor interest was already there. There were about 70 people globally, and what would become his team was around 20 of them.

The revenue picture is worth stating plainly, because it is the base everything else grew from. Grafana Labs ended 2018 at roughly $5 million and 2019 at roughly $15 million, after a slower climb from about a million in ARR. Hanna is careful to note the acceleration began before he arrived and he cannot claim credit for it.

At the time of this recording the company was around 600 people, had raised a B round and a C round, and had roughly $337 million in total funding, which was more than Crunchbase showed because two insider B rounds were only partly announced.

The clearest illustration of the change is a single decision. Early in his tenure, Hanna and the CEO sat down to debate whether to hire one or two additional salespeople, and whether even that was getting ahead of themselves. At the time of the conversation they had roughly 40 open sales roles and would have filled all of them the next day if they could.

Observability, explained for people who do not work in it

Hanna's version for a non technical audience is the useful one. Grafana Labs helps companies understand how well their technical systems are performing. Fast or slow, online or offline, working or not.

The range of what people watch is the fun part. Hobbyists monitor beehives, home labs and solar panels. Fortune 500 companies watch whether applications are performing, whether ATMs are online, and whether the Wi-Fi in a theme park is working.

Grafana itself started about seven years before this recording and focuses on visualizing time series data, meaning anything where time is one of the axes. Dashboards, charts and alerts, heavily customizable. At the time there were about 800,000 active instances and a couple of million end users.

How the open source business actually makes money

Grafana is free in both senses, no cost and open source, downloadable and modifiable, and available free in Grafana Labs' own cloud through a freemium tier.

Hanna draws a specific contrast with WordPress, which he and Michael both know well. Grafana Labs owns and controls the Grafana project rather than handing it to a foundation, and it monetizes two ways: Grafana Enterprise, a premium version with additional security and compliance features and more data sources, and a managed service. In the WordPress ecosystem, by contrast, the money is largely in managed hosting, since nobody is selling a premium WordPress.

Where the managed service earns its keep is complexity. Running a Prometheus metrics instance at scale is tedious and demands specialized expertise, so partnering is often the rational choice.

Adopt, land, expand

Grafana Labs' go to market motion has three stages, and the first one is unusual.

Adopt. Most people arrive as open source users, downloading Grafana from the website or GitHub, or starting with the free tier in the cloud. The large majority stay there permanently and never pay, which Hanna treats as fine rather than a leak. Open source software is genuinely good and you do not have to pay for it.

Land. A small percentage convert. Sometimes Grafana Labs reaches out, sometimes the user engages with webinars, events and content, and sometimes an inbound request arrives because a team is scaling and wants support or specific features. From there the sales team works alongside solutions engineering, legal, finance and internal deal support to get an offer in front of the customer.

Expand. Account executives stay involved after the sale. Grafana Labs deliberately does not split hunter and farmer roles, using a combined role instead. Customer success comes in, many customers buy professional services, and support runs around the clock. Regular business reviews follow, checking six months in whether the outcome matches what was promised during the sale. If the customer expands, they are almost certainly renewing, and the flywheel turns.

Hanna adds an important caveat to anyone reading this as a template. The low touch version, where someone signs up online and never speaks to a human, is not their model for most revenue, because the average customer their sales team handles is spending a lot of money in a high end enterprise sale. They do run a self service business, it is growing and strategically important, but the two coexist.

He also notes how often the sequence runs the other way. Zendesk did not hire its first salesperson until somewhere around $10 to $15 million in ARR, running on the support team until then. Shopify automated a great deal of it. If your product can grow that way, it is an efficient path, and enterprise sales usually gets added later when bigger customers arrive wanting custom pricing and negotiated terms.

How the sales organization is cut

This is the most directly transferable part of the episode.

Grafana Labs organizes sales regionally first. In the US, East, West and Central, each with a leader. In EMEA, North and South, with Central being added. Plus APAC. All of it reports to a head of sales, who reports to Hanna.

Two things are pulled out of that structure entirely. US federal, and channels and partners.

Within each region, roles are segmented by customer size measured in employees. Commercial covers companies under 1,000. Enterprise covers 1,000 to 5,000. Strategic covers 5,000 and above. A regional team carries a mix of all three, and accounts are assigned by where the company is based.

The reasoning behind what does and does not get its own team is the useful part. Federal is separate because everything about it is specialized: how agencies buy, how they procure and evaluate software, how budget appears, who you work with. Support is specialized too, since some federal entities and projects can only be supported by US citizens.

Financial services, despite being one of their strongest markets, is not separate. Most large financial services companies are paying customers, and JP Morgan named Grafana Labs one of its top vendors, which mattered because that firm sets technical direction others follow. But the product engagement is not fundamentally different from any other large customer. Complex buying processes and multiple stakeholders, yes, but the same product used the same way. Geography handles it informally anyway, since the biggest financial firms cluster in Lower Manhattan, so the team covering that territory naturally carries a financial services book.

Hanna's transferable point about verticals is short. One big proof point is enormously helpful, and you should leverage it as far as it will go.

The stages, and what actually changes

Hanna sketches the growth stages without pretending the boundaries are clean. There is no letter in the mail telling you that you have crossed from pre to post product market fit, and calling it requires conviction that is not obvious at the time. Grafana crossed it around when he joined, possibly before.

After that comes scaling, and their defining goal for the year was repeatability and predictability. Concretely, that means knowing how many prospects you need to talk to, what pipeline that produces, what the sales process looks like, how long it takes, and how frequently you must run it to hit a target. It took the whole year to work out. He notes that halfway through the quarter there were still plenty of balls in the air, and that this is simply what enterprise sales is.

Now the question is different again. How do you double, triple or quadruple that without everything breaking.

The honest answer is that he does not fully know. What he does know is what will be hardest. At roughly 600 people, growing headcount 70 to 80% the following year means about 500 net good hiring decisions in twelve months, spanning salespeople, engineers, VPs and probably C level executives, while continuing to run the business. As he puts it, he is not a professional recruiter and has other things to do. Making ten or fifteen people successful is a solved problem. Doing it with a hundred is a different set of challenges, and his expectation is that things will break and they will refine as they go.

Being hard on yourself, and the cost of it

One of the most candid passages is about internal culture during a year that looked, from outside, like an unbroken run of good news.

Hanna estimates the team spends 98% of its energy on what is broken and what needs to be better, and 2 or 3% on what they have accomplished. He thinks that ratio is common among successful teams and is part of why they improve. He also thinks it distorts the picture.

His description of being inside a fast growing company is one of the more reassuring things I have heard an operator say out loud. It feels like everything is broken, everything is on fire, and nothing works the way you want. He is confident the best company in the world has its own version of this. The corrective is stepping back to look at the year in aggregate, at which point the progress becomes visible again.

The operating cadence

Asked how the company runs itself, Hanna notes that much of it has changed less than you would expect given the growth, largely because he installed it when he arrived.

He implemented OKRs, the current all hands cadence, and the leadership team syncs. Most people in the company did not have formal one on ones before he started.

On OKRs he is deliberately undogmatic. The point is setting out priorities, deciding what to get done in the quarter, and executing. Goals cascade from company level down to individual contributors, everyone sets them, and they are all public, with retrospectives at the end of the quarter.

The go to market rhythm has three pieces: a look back at the previous quarter, pipeline reviews, and a forecast meeting covering in quarter deals. His verdict on the forecast meeting is the payoff of two years of refinement. The right people are in the room, decisions get made, and it is a genuinely valuable hour. Newer processes still feel clunky, and he is candid that they improve the acutely broken ones first and let the rest wait.

Hiring got the same treatment, with a defined process for who candidates speak to, how long it should take, how the decision is made and how it runs through to onboarding, because it is the highest throughput thing the company does.

The advice he would give himself

Asked what he would tell himself in September 2019, Hanna leads with a joke about gray hair, notes that nine quarters as a revenue leader at a high growth startup produces plenty of it, and then gives a serious answer.

Spend the time and energy to bring the right people onto the team, and do not cut corners. He is sympathetic about why companies do. Leaving a role open is painful, and the compromise is rational in the moment: someone is better than nobody, you needed the role filled yesterday, and a proper search takes months. His warning is that in a high growth environment companies compromise more than they should, and the effects show up later as a disruptive change you have to make anyway.

Celebrating wins when nobody is in the room

The last thread is one every distributed company will recognize. Since the start of the pandemic the company had raised roughly $300 million, grown substantially, made key hires, and closed its first million dollar deal and first three million dollar deal. Everyone was sitting alone in a home office, a hotel room or a basement.

Hanna calls it anticlimactic, and admits his best effort has sometimes been buying a bottle from the wine store down the street on the company. They celebrate in person when they can, and travel had resumed for some people.

What they do have is a Slack channel called Commercial, where the rep who won a deal posts the story and thanks the people who helped. On the day of the recording they were closing a deal more than two years in the making, and Hanna was looking forward to that post collecting several hundred emoji reactions from across the company.

The part worth keeping is what he says next. From outside it looks like a smooth run. He cannot count how many times he talked that rep back from a bad place, or how many testy Slack messages they exchanged about the same deal.

The 5 things I took away from this conversation

1. Segment by how customers buy, not by how you would like to org chart them. Federal got its own team because procurement, evaluation and even support staffing are genuinely different. Financial services did not, because a bank uses the product like everyone else. That test, does the buying or the using actually differ, is better than the usual instinct to carve verticals because they sound important.

2. Repeatability is a stage, and it takes about a year. Hanna's definition is precise: knowing how many conversations produce how much pipeline, over what cycle, at what frequency, to hit a number. Most teams claim predictability long before they have measured this. Naming it as the explicit goal for a full year is the discipline.

3. One proof point unlocks a vertical. The JP Morgan award did more than a campaign would have, because that firm sets the pattern others follow. If you are trying to enter a market, the whole strategy is landing one reference everyone else watches.

4. Do not split hunter and farmer too early. Keeping account executives involved after the sale is a deliberate choice at a company where deals are large and expansion is the growth engine. Handing customers off at signature optimizes for the wrong moment.

5. The 98/2 ratio is real, and it lies to you. Spending nearly all your energy on what is broken is why good teams improve, and it also means everyone inside a successful company thinks it is on fire. Scheduling the look back is not a morale exercise. It is a correction to a systematically distorted view.

FAQ

What is revenue operations in practice at a growth stage company? At Grafana Labs it covers the machinery underneath all customer facing teams: how the sales organization is segmented, how pipeline is reviewed and forecast, what the sales process looks like and how long it takes, how customers are supported after signature, and how all of that is measured. Hanna owns it alongside the revenue teams themselves.

How should a sales organization be segmented? Grafana Labs segments regionally first, with US East, West and Central plus EMEA and APAC teams, then by customer size within each region, using employee count. Under 1,000 is commercial, 1,000 to 5,000 is enterprise, and above 5,000 is strategic. Federal and channel businesses are pulled out entirely because the buying process differs fundamentally.

When should a company hire its first salesperson? Later than many assume. Hanna notes Zendesk ran on its support team until roughly $10 to $15 million in ARR. If your product can be bought self serve, that is an efficient way to grow, and the enterprise sales motion typically gets added when larger customers start asking for custom pricing and negotiated terms.

What does predictable pipeline actually require? Knowing the full chain: how many prospects you speak to, what pipeline that generates, what the sales process involves, how long it takes end to end, and at what frequency you need to run it to hit the target. Hanna says it took his team a full year to establish, and it remained the central goal for that year.

How do you scale a startup business past product market fit? Hanna's sequence is find product market fit, then establish repeatability and predictability in go to market, then work out how to multiply that without it collapsing. He is direct that the third stage is unsolved for them, that hiring at volume is the binding constraint, and that things will break and be repaired continuously.

Also mentioned

• Grafana Labs, Grafana Enterprise and the managed cloud offering
• Prometheus, whose operational complexity at scale drives managed service demand
• Zendesk and Shopify, cited as companies that grew far on self serve before building enterprise sales
• OKRs, implemented company wide and kept public, with quarterly retrospectives
• AppDynamics, acquired by Cisco reportedly just before its planned IPO, Hanna's favorite illustration that anything can happen
• Stripe and Databricks, examples of companies staying private longer on abundant private capital
• Part one of this conversation, on scaling the organization from 70 to 500 people

Listen to the full episode

Doug Hanna on Between Two COO's, part 2

Between Two COO's is hosted by Michael Koenig. Subscribe on Apple Podcasts, Spotify, or wherever you listen.