Scaling Support and Customer Success

Channel and Org Structure Overviews

Dealing with Traffic Surges

Zapier deployed several tactics in the past to get out of backlogs:

• Prioritize high-value customers 
• During a backlog, temporarily dedicate agents to specific topics to avoid context switching e.g. in one of their backlogs, they had one agent handle all Meta integration issues for the period
• Run bug bash days - when you've already missed SLAs and customers are frustrated anyway, abandon first-in-first-out for high-energy sprint days to clear demotivating backlogs
• Involve the wider company. 
  
  • Monzo also did this in 2018. It helped clear their backlog, boost morale and it also gave EPD team members first hand insight into support challenges; this increased empathy and led to a culture of product teams considering support implications when building e.g. creating scalable support flows and customer view data pipelines before shipping
  • Meanwhile, a transparent post mortem blog post created goodwill with users 

During volume spikes, Zapier and Vercel emphasised one-to-many response and expectation setting:

• Set clear expectations - communicate realistic turnaround times upfront when tickets are submitted; use mass communications (email blasts, help page banners) for widespread issues
• Surface common patterns - identify recurring questions and address them at entry points

In Batch 1, Karen Peacock (former CEO of Intercom) shared another tactic. Intercom prioritized customers who were in their first two weeks, aiming to answer their questions rapidly with the belief that early interactions are critical to building trust. Once customers are more established, they've built enough trust to tolerate longer response time, but initial touchpoints require more immediate attention.

Hiring and Team

Support team evolution

Lessons from Monzo:

• When starting the support function, Monzo staffed the frontline with overqualified generalists (to give you a sense of their level, these folk would eventually become Ops / Engineering leaders)
• This enabled Monzo to process high volume regardless of complexity, but also have smart systems thinkers start to put in place more scalable structures e.g. forecasting and single customer view systems that are still in use 9 years later today
• As Monzo grew, generalists graduated to become support specialists while the company hired typical support agents as frontliners. This created a scalable flywheel of graduation and knowledge transfer. 
  
  • Specialists a) trained new frontliners and b) designed scalable training programs including "Badges" - specialized curricula that frontliners could complete to earn the right to handle specialised query types (like Fraud or Lending) for additional pay. 
  • These Badges eventually transferred to Product teams, who became responsible for creating training and ensuring adequate customer service for new products they shipped
• Despite this structure, Monzo still hit severe support backlogs as growth accelerated. The team lacked support capacity. To get out of this, the whole company did support for a few days (mentioned above) and the team prioritized aggressive hiring through external support recruiters, referral bonuses, and support job fairs – driving support team applications from 20 to 80 daily. 
• They also invested heavily in hiring and onboarding infrastructure to accommodate 25+ new hires per week. For example, Monzo ran assessment days where groups of applicants would be assessed at one time rather than in individual interviews.  
• At maturity, Monzo has now outsourced frontline support to a 3rd party BPO called Teleperformance, with specialists and Badged agents still handling more sensitive and complex issues in-house

Lessons from Zapier:

• The transition from generalist ‘do anything’ support culture to specialised support roles with KPIs/SLAs can be tough. The Zapier Director of Support notes that 25% attrition is normal as early stage support people now have to work to KPIs and this goes against their preferred way of working
• In terms of KPIs, Zapier track: first reply time, replies to resolution, full resolution time, CSAT scores 
• Re: support people management:
  
  • Zapier found that attitude in support can be as important as KPIs. Support is an ever evolving space and is never ‘solved’; meanwhile bad attitudes are infectious within support teams. An agent with a positive attitude and growth mindset hitting 30 tickets a day is a better agent than one hitting 40 tickets a day but complaining a lot and spreading negativity in this ‘never done’ role. Zapier recommends weeding these people out early. 
  • Zapier recommend having established graduation paths for top performers e.g. progression to team leads, TAMs, junior eng roles to keep up morale
• Zapier recommends making the transition to KPI-driven support asap. The longer you leave it, the bigger the team gets, the longer habits are formed, the harder it becomes. 

Clients who are more technical/developer oriented want success support from a different type of person than a traditional enterprise, vice versa. Nuanced hiring and incentivisation for this is key. 

• Retool hired a typical sales leader to run Success. In retrospect, this was the wrong move; the leader instilled sales-based incentives and hired similar folk to them. This created a success team that was too ‘salesy’ for a core developer ICP. Indeed, all Retool’s smart, technical success people ended up leaving because of this culture, and developer clients felt overly 'sold to' causing customer churn 
• Vercel had similar issues with wrong early hires. This set them back years as leaders brought in similar folk to themselves. Both Vercel interviewees noted that success requires hiring different personality types for enterprise vs. technical support and deliberate incentive design nuanced across sales vs enablement
• The Zapier Director concurred and suggested turning internally where possible: "Make your best support agents Technical Account Managers since they understand and care about the product deeply."

Managing Top of Funnel

Meet your customer where they are with targeted content. Zapier analyzes support ticket patterns to identify usability friction points, then deploys targeted help articles to address them. They found articles around ‘Getting Started’ were the highest leverage. Zapier emphasised ‘meeting your customer where they are’ – for Zapier, since most users go to Google / Reddit rather than in-product docs for support, they created SEO-optimized articles and posts that serve dual purposes as support content & blogs.

Ensure help content is tagged to customer language, rather than internal product terminology. Zapier discovered gaps between internal feature names and customer search terms, requiring content retagging to improve content discoverability. This helped deflect inbound support. For example, whilst Zapier internally refers to “Memory” in its Agent product, they found customers often refer to it as ‘accuracy’ or ‘remembering’. 

GPT-summary style support bots are table stakes, should include internal AND external content, and should be monitored, managed and improved like any product feature. 

• The Zapier Director of Support advocates replacing decision tree support bots with AI enabled bots and providing GPT-like summaries at the end
• Provide bots access to ingest both internal documentation and external resources (GitHub, Discord, Reddit communities) where users may already be solving problems
• You should have continuous monitoring and improvement of bot performance and treat the bot as you would any product feature. Invest in tracking success rates because you need to continuously improve the bot. Where bot improvements don’t move the needle on success rates, this is a good signal that you have a product problem.

Invest in community support…and shout about it. Vercel staffed forums with technical team members responding to key threads. They found creating content around examples where community posts led to product wide solutions helped them market their technical expertise and developer experience.

Platform Investments

Tooling investment. The Zapier CEO mentioned that internal tooling (tactically, internal chat bots and recommended hot links to knowledge bases) increased tickets per agent by 2x. Zapier also uses Scribe AI and Synthesia to help generate text and video help content. 

Invest in ticket metadata for process improvement and for agent routing. To be able to map pain points and route to specialists efficiently, Zapier recommends using structured support forms, or at least having a categorisation engine analyzing inbound emails. They also have built routing so that enterprise queries get routed to designated account managers regardless of contact channel, noting that excellence means ‘meeting the customer where they do their work and want to be met”.

Dedicated floater support makes bug fixes a priority. Zapier recommends maintaining dedicated floating engineer capacity whose priority it is to fix genuine bugs for enterprises as they come in.

Support <> Sales Relationship

Sales, handoff and onboarding

Success can only be as good as your pre-sales process, and this needs to be constantly evaluated as your product and ICP evolve. 

The Notion CSM emphasizes that "you can have the best CSMs in the world - but if the deal wasn't properly qualified to an ICP and/or oversold, you are screwed down the line." Zapier also warned against this: “don't let sales over-promise - set clear service boundaries upfront to avoid renewal disappointments”. 

Tactics: 

• Notion's non-ICPs using single features resulted in churn when point solution alternatives emerged. They recommend enforcing Sales to do ICP acquisition rather than broad volume with top down repetition and goal setting e.g. goal on X meetings within the ICP category rather than sheer volume. 
• The Notion CSM emphasizes re-evaluating ICP as product and buyer behaviour changes over time, and re-training your sales team in lock step with these changes by refining pitches for each new use case. They work in tandem with product marketing teams to make sure messaging is consistent across teams and the org

Customers having to re-explain themselves post handoff needs to be avoided (plus some tactics for building trust on Day 1). 

The Notion CSM identifies poor handoffs as creating "the worst outcome" of customers having to re-explain themselves. It is important that roles and responsibilities are defined, customer priorities are shared and everyone in the GTM org can see a single customer view in one system. The Director of Support at Zapier re-emphasised this. 

Tactics:

• Zapier had a template that CSM required Sales to fill out. Fields included: reasons for sign up, success after 12 months looks like X, customers’ tech stack, specific requirements e.g. security policies. They also have a handover meeting. Where possible, Zapier matches Sales people to the same CSMs so they build good ways of working together over time. At Zapier a deal does not move from ‘Sold’ to ‘Secured’ until CSM signs off on the handover
• Notion implements similar systems requiring complete information transfer before deal finalization e.g. north star goals, why the customer is using you. Like at Zapier, sales teams do not receive deal credit until all customer data is properly documented, and CSMs have validated completeness
• To help address the problem that CSMs don’t know much about the customer on Day 1…Zapier have Sales hold back one valuable suggestion/insight for CSMs to share on Day 1 to instantly build trust with the client 
• To address the same problem of building trust, Notion have the CSMs share relevant personal stories and results from similar clients during all kick off calls to build trust

At onboarding you should mutually agree explicit 90 day goals and joint owners to create structure and buy in. Notion found "the first 90 days of a customers' experience dictate the entire relationship direction." Tactics:

• In kick off calls, collaboratively workshop a 90-day north star success metric to generate buy in e.g. x% adoption, y number of features used, z external integrations connected
• The CSM should then go away and revert with a detailed project plan working backward from agreed north star metrics and assign multiple stakeholders as owners on both vendor and client sides to create multiple points of engagement and internal champions. Do weekly stand ups.

Post-onboarding enterprise management

Value demonstration. The Notion CSM advises beginning the post-onboarding phase by reviewing achieved 90-day metrics to demonstrate that you do what you say you will and maintain trust momentum.

Enterprise support thresholds and ratios need to be designed to fit your specific business. 

• Vercel gives high touch support to top tier i.e. those crossing a specific revenue threshold ($300k+), strategic logo value, and company size (~5000 employees), using contract renewal as a milestone for any re-bucketing between tiers. 
  
  • They originally structured teams with CSMs managing 5 accounts at the top tier end of enterprise. In Aug 2025, their new COO restructured their teams and let go of 75% of CSMs citing duplicated work and IPO preparation. Now, 9 CSMs split the top tier of 220 enterprise accounts, and AEs cover the rest of their smaller clients (~50 each), with a new startups team covering early stage startup business. 
• Notion has a more nuanced view. They developed lifecycle management programs around distinct personas e.g. you can be big and lucrative, but you may also be technically capable and not need much support. 
• Notion segments customers across: size, spend, seats, feature breadth, and tech sophistication. Different personas receive tailored support programs e.g. small companies who are tech forward may not get much support post-onboarding, but Notion would engage them with new feature launches and support around those. Larger, less tech savvy enterprises will get more of a white glove service and multiple touch points between EPD teams. These programs require re-evaluation as your product and your customer’s organizations/familiarity evolve. Notion tends to re-evaluate this at QBRs.

‍