Support

Fifteen years ago, customer support at most ecommerce companies looked roughly the same: you sent an email, you waited, you hoped. Larger retailers operated phone lines staffed during business hours. The very largest had call centers in lower-cost regions overseas. Behind the scenes, support teams used ticketing systems — Zendesk, founded in 2007, was already the dominant player by 2011 — to track each customer interaction as a thread, assigning it to an agent and watching the queue grow.

The model had obvious problems. Hold times stretched into hours during busy periods. Email response times were typically measured in days, not minutes. Customers had no visibility into the queue and no expectation of speed. For ecommerce specifically, support frequently lagged the rest of the experience: you could complete a checkout in thirty seconds, but resolving a delivery problem might take two weeks. In gaming, the disparity was even sharper. World of Warcraft’s in-game ticket system at its peak ran response times of multiple days. Steam Support became a recurring meme — "Steam Support replied within the year" was a punchline, not an exaggeration.

The first major shift began around 2011, when consumer-grade live chat moved from a niche enterprise tool to a near-universal ecommerce feature. Companies like Intercom, founded in 2011, and Olark, founded in 2009, made it cheap and easy for any online merchant to drop a chat widget in the corner of their site. Within five years, live chat had become an expected feature of any serious ecommerce store. Customers no longer wanted to wait for an email reply when a real person was sitting on the other side of a webpage.

The chat-first model rewired support economics. Average response times dropped from hours to minutes for chat-handled queries. Conversion rates rose noticeably for stores that offered live chat — a customer about to abandon a cart could be saved by a thirty-second exchange. For gaming, live chat appeared first in the major publisher portals. Riot Games and Blizzard both invested heavily in in-game support tools and live agent escalation paths; smaller studios followed once Discord and third-party widgets made the cost of a chat channel trivial.

The next shift was less visible and arguably more important. Around 2013–2014, support tools added structured knowledge-base modules — searchable help centers where customers could find answers without ever opening a ticket. Zendesk’s Help Center, Help Scout’s Docs, and similar products from Intercom and Freshdesk all launched in this window. The strategic logic was simple: every ticket prevented by self-service was a ticket the support team did not have to handle.

Done well, self-service doubled or tripled the effective capacity of support teams. Done badly, it became a maze of stale articles that frustrated customers more than no help at all. The ecommerce companies that invested seriously in their knowledge bases — clear writing, regular updates, search-engine-friendly structure — saw measurable improvements in customer satisfaction even as ticket volumes fell. The gaming industry mirrored the trend. Steam’s community support pages, the Blizzard Battle.net knowledge base, and the per-game FAQ pages that proliferated across publisher sites all date to this period. Players still complained about wait times, but more of them were finding their answers without needing to wait at all.

By the mid-2010s, customers were no longer asking support questions through a single channel. They were asking on Twitter, on Facebook, in Reddit threads, in app store reviews, and in private messages on whatever platform they happened to use that day. Companies that ignored these channels lost goodwill quickly. Companies that staffed them well — JetBlue’s responsive Twitter team became a famous case study — turned support into a public marketing function.

The challenge was no longer answering the question. It was answering it consistently, across every channel, without forcing the customer to repeat themselves. The omnichannel response, which dominated support tooling from roughly 2017 onward, involved unified inboxes that pulled email, chat, Twitter, Facebook, SMS, and in-app messages into a single agent view. Intercom and Zendesk both rebuilt their products around the unified-inbox concept. For gaming, Twitter became the de facto status page; outages on League of Legends, Fortnite, or Steam were typically acknowledged on Twitter before the official status page caught up. Players learned to check both, and the smarter publishers staffed both.

The COVID-19 pandemic of 2020 compressed five years of support evolution into eighteen months. Phone centers shut overnight. Support teams went remote, often for the first time. Ecommerce volume surged across nearly every category — at one point, Amazon was hiring 100,000 new workers in a single month to handle the demand. Self-service usage spiked because customers had no other realistic option, and companies that had under-invested in their knowledge bases discovered they could not catch up fast enough.

Gaming was hit harder than most categories because it was simultaneously a lockdown beneficiary and a live-service product. New players flooded games like Animal Crossing, Fortnite, and Valorant, and existing live-service titles saw concurrency records broken nearly every week. Support queues stretched. Account recovery, payment disputes, and connection issues all spiked together. The publishers that came through the period best had already invested in self-service and automation; the ones that had not were forced to do it under emergency conditions. By the time the surge stabilized in 2022, the support function across both ecommerce and gaming looked permanently different from 2019.

The most recent shift began in late 2022, when large language models became commercially available. By 2024, every major support platform had launched some form of AI agent capable of handling first-line queries. Klarna, the Swedish payments company, publicly reported that an AI assistant was handling roughly two-thirds of its customer service chat volume within its first month of deployment. Intercom rebuilt its entire product around an AI agent. Zendesk pivoted toward what it called a "resolution platform." Smaller ecommerce companies followed within months, often by integrating off-the-shelf AI agents into their existing help-desk software.

The results were mixed. Done well, AI support deflected routine queries, summarized conversations for human agents, and surfaced relevant knowledge-base articles in real time. Done poorly, it produced confidently wrong answers and frustrated customers who wanted a human. Gaming publishers were among the first major adopters and among the first to learn that some queries — account recovery, payment disputes, hardware troubleshooting — were better escalated immediately to a human agent. For Cosmo Strategy Guides readers, the practical implication is the same as it has been for fifteen years: when something goes wrong with an online purchase or an online game, the speed and quality of the support experience increasingly determines whether you stay a customer. The technology has changed. The principle has not.