This is the most counterintuitive advice in this entire blog, and it needs careful context before you misapply it.

Basecamp offers only email-based support, with no live chat option. Their team reports that this deliberate choice reduced total volume, improved resolution quality because agents had time to think, and improved CSAT because responses were more thorough.

A SaaS company added a "Have you checked our help center?" page between the "Contact Us" link and the actual contact form. 25% of visitors found their answer in the help center and never submitted a ticket. Total support volume dropped. The tickets that came through were genuinely complex.

A small amount of intentional friction, placed between the customer and the support team, filters out queries that could be self-served and lets through queries that genuinely need human attention.

The Important Caveats

This is not "hide the contact button." Companies that bury support behind five menu levels, require logins to access help, or replace their contact form with a chatbot that refuses to connect to a human are doing something different. They're blocking support. That destroys trust and drives customers away.

This is also not appropriate for all customers. Enterprise accounts, VIP customers, customers in crisis, and customers with billing problems should have frictionless access to support. Adding friction for a customer who was just charged $500 incorrectly is a terrible idea.

The friction should be helpful friction. It should redirect the customer to self-service resources that might actually solve their problem, not to a wall that prevents them from getting help.

The Theory: Effort Filtering

Most support queues contain a mix of ticket types:

Type A (30 to 40%): questions that have standard answers available in documentation, FAQ, or self-service. "What are your hours?" "How do I reset my password?" "What's your return policy?"

Type B (40 to 50%): genuine issues that need human attention. Billing errors, bug reports, complex how-to questions, complaints, account-specific problems.

Type C (10 to 20%): requests that are urgent or emotional and need immediate attention. Service outages, security concerns, angry customers, bereavement requests.

If you make it equally easy to submit a ticket for all three types, you get all three types at equal volume. But Type A tickets don't need a human. They need a help article. Spending agent time on "what are your hours?" is a waste that slows down the Type B and Type C tickets that actually need attention.

A small effort filter (a help center page between "Contact Us" and the actual form) catches Type A tickets. The customer sees a search bar and the top FAQ articles. If their question is answered there, they're done. If not, they click through to the contact form.

The filter doesn't stop Type B and Type C tickets. A customer with a billing error won't be satisfied by an FAQ article. They'll click through to the form. A customer in crisis will click through immediately. The filter only stops the people who could have self-served.

How to Implement Without Being Evil

The friction must be helpful, not obstructive. The intermediate page should show genuinely useful content, not a wall of irrelevant links. AI can help here: if Supp's widget shows the top 5 help articles relevant to what the customer typed (before routing to a human), the customer gets immediate help for simple questions and smooth escalation for everything else.

The path to a human must always be visible. "Can't find what you need? Contact us" should be on every self-service page. Not hidden. Not behind another click. Right there. The customer should never feel trapped.

Some customers should bypass the filter entirely. VIP accounts, customers with active billing disputes, customers who've already submitted a ticket about the same issue, and customers flagged as high churn risk should go straight to a human. The filter is for first-time, non-urgent queries only.

Monitor the impact. If self-service deflection is high and CSAT stays stable, the friction is working. If CSAT drops, you've added too much friction or the self-service content isn't good enough.

The Numbers

A well-designed effort filter typically reduces ticket volume by 15 to 30%. The remaining tickets are, on average, more complex and more valuable (they represent real issues that need real help). Agent handle time per ticket goes up slightly (because the easy tickets are gone), but total agent workload goes down.

Response times for complex tickets improve because the queue is shorter. CSAT on complex tickets improves because agents have more time per ticket.

The customers who self-served got faster answers than they would have waiting in the ticket queue. The customers who submitted tickets got faster, better responses because the queue is shorter. Everyone benefits.

When This Backfires

If your self-service content is bad, the effort filter becomes a frustration wall. The customer sees the help center page, searches for their issue, finds nothing useful, and then has to submit a ticket anyway. They're now more annoyed than if they'd been able to submit the ticket directly.

The filter only works if the self-service content is good. If your help center has 10 outdated articles that cover 20% of your ticket categories, adding a filter in front of it makes things worse.

Fix the content first. Then add the filter.

If certain customer segments have low technical literacy or high urgency (elderly users, medical customers, financial customers), a filter adds cognitive load that these customers can't afford. Know your audience. Apply friction selectively.

This isn't a universal best practice. It's a targeted optimization for companies with good self-service content, moderate to high ticket volume, and a significant percentage of tickets that could be self-served. If that's you, a small amount of helpful friction improves everyone's experience. If that's not you, fix the fundamentals first.