Coresee

BlogGuidesAgencies & studios

Browser-First Collaboration for Agencies: What Belongs in the Browser and What Doesn't

9 min readNovember 9, 2026

Summarize

TL;DR

Going browser-first is a decision about how an agency works with people it doesn't employ, on devices it doesn't manage. Review, approvals, client sessions and coordination belong in a browser. Editing, compositing, grading and audio don't. Agencies that write that boundary down end up with a smaller stack that people actually use. The weak points to plan for are dropped sessions, offline work and offboarding, and for a national team, a shared working day of five hours at most.

Why do agencies go browser-first?

Because of who agency work involves. Freelancers, production partners, post houses and client stakeholders rotate through every campaign, and almost none of them use machines the agency administers. A browser is the one environment every one of them already has, whatever their employer, device or IT policy.

Distributed working is also a settled condition. Telework is concentrated in the sectors where agencies and their clients sit: 47.2 percent of workers in information and 41.8 percent in professional and business services teleworked in June 2026.1

The browser also makes live sessions possible without installs. The W3C publishes WebRTC as a Recommendation defining the browser interfaces that carry real-time audio and video between browsers and devices,2 which is why a client can join a live presentation from a link.

Which agency work belongs in a browser?

Review and approval, without question. Viewing work, commenting, comparing versions and recording a decision all work well in a browser, and these are the workflows with the most outside participants, which is where the browser helps most.

Live client sessions. A client who can join from a link, on their own laptop, joins on time. An installation request to a client’s IT team can take longer than the review cycle.

Coordination. Scheduling, briefs, status and asset handoff are text and file work. They gain nothing from a native application, and an installed tool shuts outsiders out.

Not creative production. Editing, compositing, grading, design and audio depend on local hardware, plug-ins and file access that browsers deliberately restrict. Put the boundary here from the start. Trying to move production into a browser is where most browser-first initiatives fail.

Where agency work belongs

Keep in native applications

  • Editing and compositing
  • Color grading and finishing on calibrated displays
  • Design work that depends on local fonts, plug-ins and files
  • Audio mixing
  • Any work that has to continue with no connection

Move to the browser

  • Review, comments and approvals
  • Live client presentations
  • Sessions with freelancers and production partners
  • Scheduling, briefs and status
  • Asset handoff to people outside the agency
The dividing line is who takes part. Work that involves people outside the agency belongs in the browser. Work that depends on local hardware stays in native applications.

Most agencies end up hybrid by design: browser-first for anything that involves people outside the agency, native applications for production inside it. That isn’t a compromise. Final color, the audio mix and detailed compositing decisions need controlled hardware, and sending them to native tools is the right call.

Why does the browser work as common ground?

Agency device estates are mixed. Creative teams work on one platform, finance and account teams often on another, clients present from tablets and phones, and freelancers bring whatever they own. No single native application covers that spread without exceptions.

The agency can’t install software on most of those machines. Installing on a freelancer’s laptop or a client’s corporate device is either impossible or slow, and slow is as bad as impossible when the review is tomorrow.

It makes brief participation cheap. When adding someone costs almost nothing, agencies bring in the specialist, the regional stakeholder or the junior team member who should be in the room, and the work improves for reasons that have nothing to do with software.

And client IT is easier to work with. Corporate networks still restrict browsers, but a browser policy question is narrower and faster to answer than a software installation approval that the client has no reason to prioritize.

How should sign-on, guests and offboarding work?

A browser-first stack tends to mean more separate services, and without shared sign-on that becomes a pile of passwords that staff manage badly and the agency can’t audit.

Federation is the mechanism to ask for. NIST describes it as letting one credential service provider supply authentication attributes to many separately administered relying parties.3 In agency terms, staff sign in once against the agency directory and reach every tool from there.

Federation matters most for offboarding. Disabling one directory account should remove access everywhere. An agency that removes accounts tool by tool will miss one, usually for a freelancer who was added in a hurry during a busy month.

Clients and freelancers usually can’t sign in through the agency’s directory, so the guest path needs its own evaluation. Check how a guest is invited, what they can see, and how their access ends, separately from the enterprise identity features.

What happens when a session drops?

Browser sessions end for reasons a native application would survive: a closed tab, a laptop going to sleep, a browser update, or a tab the browser suspends to save memory. Each interruption is small, but together they decide whether people trust the environment.

Ask during evaluation, and test it yourself:

  • When a participant rejoins, do they land where they left off, or back at the start?
  • Does an interrupted upload resume, or start over?
  • Does a session survive the machine sleeping for ten minutes, which is what happens when someone steps out of a long review?

Then set some habits. Build breaks into long sessions, ask participants to keep the session tab in the foreground while uploading, and avoid scheduling critical reviews right after a known browser update.

What happens when the connection fails?

This is the real weakness of browser-first work. Work that lives entirely in a browser is unavailable when the connection is, which matters for staff who travel, present on location or work on unreliable connections.

The web platform does offer a mechanism. The W3C Service Workers specification provides a request and response store similar in design to the HTTP cache, which makes offline-enabled web applications easier to build.4 Whether a given product uses it is a separate question, so ask what a participant can do with no connection, whether material they already loaded stays viewable, and whether anything they create offline syncs when they reconnect.

Where the answer is limited, plan around it. Keep an offline copy of anything you’ll present where connectivity is uncertain, and make that standard practice for on-location work.

What changes when the team spans the country?

A national agency hits three things a single-city agency doesn’t: a short shared working day, a wide spread of connection quality, and clients who notice when different offices run sessions differently.

Start with the clock. With everyone working 9 to 5 local time, New York and Los Angeles share five hours, from noon to 5 p.m. Eastern. Leave out the first and last hour of each day and the realistic window for a live review is three or four hours, and it shrinks further when Alaska, Hawaii or an international partner is involved.

The shared working day across U.S. time zones

0611 hours after 9 a.m. ET

Eastern

9 a.m. to 5 p.m. ET

Central

10 a.m. to 6 p.m. ET

Mountain

11 a.m. to 7 p.m. ET

Pacific

Noon to 8 p.m. ET

Everyone working

Noon to 5 p.m. ET, five hours

Four time zones working 9 to 5 local time overlap for five hours, from noon to 5 p.m. Eastern. Leave out the first and last hour and three to four hours remain for live sessions.

Spend those hours on decisions that need conversation, and let notes on fixed material be written whenever suits each reviewer. Fix a standing review slot in the middle of the overlap so nobody negotiates calendars each week. And decide who absorbs the inconvenience: defaulting to headquarters time quietly puts the cost on whichever coast isn’t home.

Connection quality varies by place, not by employer. The FCC’s National Broadband Map lists the internet services providers report at individual addresses, including the maximum advertised download and upload speeds.5 Before a remote hire or a presentation from an unfamiliar location, check the upload figure as well as the download, since a presenter sends video as well as receiving it. Then plan sessions around the weakest connection likely to be in the room, because a national team always has one.

Client networks vary too. A platform that runs cleanly for one client may be blocked by another’s filtering without anything changing on the agency side. Ask your vendor for a technical requirements sheet that names the domains and ports the platform uses, so a client’s IT team can act on it quickly. Keep a shared note of each client’s constraints and fixes, and run a short connection check before the first real session with any new client.

Finally, decide centrally what the client sees: how invitations arrive, what joining looks like and how feedback is captured. Make that standard setup the fastest option for every office, give it one owner, and now and then join another office’s session as an outside participant to see what a client sees.

Does going browser-first mean more tools?

At first, usually yes. Browser delivery makes adopting a service easy, so teams add tools for specific needs and the stack grows. The upside is that each tool is just as easy to drop, whereas installed software tends to linger long after it stopped being useful.

Governance decides which way it goes. Agencies with a named owner for the collaboration stack, shared sign-on and a simple way to propose a new tool tend to consolidate over time. Agencies without them accumulate.

What should an agency measure?

  • Time to first successful join for an outside participant. It captures most of what browser delivery is meant to deliver, and it worsens noticeably when something in the stack changes.
  • Sessions that start late for access reasons. Agencies rarely track this and usually underestimate it.
  • Offboarding completeness. Each month, pick a departed freelancer at random and check that their access is gone from every service. This is the risk a browser-first stack builds up fastest.
  • Not login counts. Usage figures reward the tools people open, not the ones that save them work.
FactorBrowser-first stackNative-application stack
Outside participantsAnyone can join quickly, on any deviceEach outsider may need an install or an exception
Creative productionPoor fit: local hardware and plug-ins are restrictedStrong fit: the tools are built for the work
IdentityNeeds federated sign-on, or passwords multiplyFewer services, but access is often tied to specific machines
Session reliabilityMore fragile: closed tabs, sleep and browser updates end sessionsMore resilient in long working sessions
Offline workDepends on whether the product implements itUsually available by default
Tool count over timeTends to grow, since adoption is easyStable, but harder to change
Best fitReview, client sessions, coordination and approvalsEditing, finishing, design and audio
A browser-first collaboration stack compared with a native-application stack

How to roll it out

  1. Write the boundary down. List which workflows run in a browser and which stay native, and circulate it before anything moves.
  2. Set up federated sign-on early, for the tools that support it. Connect them to the agency directory before the stack grows, so one disabled account removes access everywhere it can. For tools that don’t, keep a list of who has access and remove people as part of offboarding.
  3. Evaluate the guest path on its own. Invite a test guest from outside the agency, on an unmanaged device and an outside network, and time how long it takes them to join.
  4. Test failure before rollout. Close a tab mid-session, let a machine sleep, interrupt an upload and disconnect entirely. None of this shows up in a demo.
  5. Map your overlap and fix a cadence. If the team spans time zones, put standing review slots inside the shared hours.
  6. Measure join time and offboarding monthly. Those two numbers show the benefit and the risk better than usage statistics.

Coresee makes a persistent, browser-based video collaboration workspace, one of the tool categories this guide describes.

References

  1. U.S. Bureau of Labor Statistics. “52.5 Percent of Workers in Financial Activities Teleworked in June 2026.” The Economics Daily. 2026. bls.gov/opub/ted/2026/52-5-percent-of-workers-in-financial-activities-teleworked-in-june-2026.htm
  2. World Wide Web Consortium. “WebRTC: Real-Time Communication in Browsers.” W3C Recommendation. March 13, 2025. w3.org/TR/webrtc
  3. National Institute of Standards and Technology. “SP 800-63C-4, Digital Identity Guidelines: Federation and Assertions.” NIST Computer Security Resource Center. July 2025. csrc.nist.gov/pubs/sp/800/63/c/4/final
  4. World Wide Web Consortium. “Service Workers.” W3C Candidate Recommendation Draft. Accessed November 6, 2026. w3.org/TR/service-workers
  5. Federal Communications Commission. “How to Submit an Availability Challenge.” FCC Broadband Data Collection Help Center. Updated May 15, 2025. help.bdc.fcc.gov/hc/en-us/articles/10476040597787-How-to-Use-the-National-Broadband-Map
Found this useful? Share it:XLinkedInEmail

Published November 9, 2026

Frequently asked questions

People keep losing their place when a session drops. Is that normal?

It’s common, and partly avoidable. Browser sessions end for reasons application sessions don’t, such as closed tabs, sleeping machines and browser updates, so some interruption is inherent. Check whether your platform returns a participant to where they left off when they rejoin, because implementations vary and demos rarely show it. Then build breaks into long reviews, keep the session tab in the foreground during uploads, and avoid critical sessions right after a browser update.

We moved everything to the browser and our editors are unhappy. What went wrong?

The boundary was almost certainly drawn in the wrong place. Browsers deliberately restrict local hardware and file access, which is right for a sandbox and unworkable for editing, compositing or grading. Move creative production back to native applications and keep the browser for review, client sessions and coordination. Browser-first is about including people the agency doesn’t manage, not about getting rid of installed software.

Our West Coast team says every review lands in their early morning. How do we fix it?

Treat it as arithmetic, not preference. Coast to coast, a 9-to-5 day overlaps for five hours at most, and defaulting to headquarters time puts the cost on one coast permanently. Set standing slots in the middle of the overlap, rotate any session that has to sit at the edge, and keep live time for decisions that need conversation. Where a client sets the timing, say so, so it doesn’t look like an internal choice.

More from the Coresee blog

View all articles

GuidesNovember 9, 2026

How to Set Up a Persistent Video Workspace for a Post Facility

GuidesNovember 9, 2026

Choosing an Enterprise Live Event Platform: Latency, Scale and Production Control

GuidesNovember 9, 2026

How Broadcast Engineers Should Evaluate a Live Streaming Platform