The phrase "client portal" stopped meaning anything around 2019. By then, every project management tool, every services-management platform, and every CRM with an agency vertical had shipped one. The features were similar. The interfaces were similar. The promise was similar: give your client a place to log in.
The problem with all of them, then and now, is that they were built for the wrong seat.
Two waves of "client portal" software have shipped in the last seven years. Both solved problems that were real. Neither solved the problem the client actually has.
Wave 1: file-share with login boxes (2018-2022)
The "give the client a login" generation.
The first wave was driven by an obvious agency pain: clients kept asking for status updates, and the agency was tired of answering. The fix was a portal: log in here, see what's in flight, see the deliverables, pay your invoice.
The pattern across a dozen vendors was nearly identical: a project board, a file repository, a comment thread, a billing module. The agency configured it. The client logged in occasionally. The portal got used heavily for the first 30 days of an engagement, then drifted into "I'll just email" mode for everyone except the project manager.
The Wave 1 tools weren't bad software. They were good software for the wrong job. The job they were built for was internal project management with a client-facing veneer. The job the client actually had was decision-making, and Wave 1 didn't help with that. Logging in to see a Kanban board doesn't help you decide whether to keep your agency next quarter.
By 2022, every Wave 1 tool was reporting the same usage pattern: heavy adoption by agency PMs, light adoption by agency owners, near-zero adoption by clients past month two. The portals existed. The seat at the table was still empty.
Wave 2: internal-team AI copilots wearing a client-facing badge (2023-2025)
The "we added AI" generation.
By 2023, every Wave 1 vendor had bolted an AI feature onto their existing platform. ClickUp Brain. Notion's Ask Notion. Slack's AI search. Productive's AI Search. Wrike's AI highlights. HoneyBook's drafting AI.
The pitch was always the same: ask any question of your workspace and get an answer. Type "what's the status of the homepage redesign" and the AI summarizes the project board.
This is useful. It's also, structurally, an internal-team tool. The "workspace" the AI knows about is the agency's workspace. The data it answers questions of is the agency's project board. The user it serves first is the agency owner who wants to skip the meeting where someone reads the board out loud.
This is what's missed in most product reviews of Wave 2. ClickUp Brain doesn't help your client decide whether the homepage is good. It helps your project manager decide whether the homepage is on schedule. Different question. Different seat.
The clients who got Wave 2 features pitched to them as "now your portal has AI" almost universally found that the AI didn't answer the questions they actually had. Not because the AI was bad. Because the AI was answering questions of the agency's data, not questions of the client's business.
If a CMO asks ClickUp Brain "is my marketing money producing return," ClickUp Brain says "I don't have access to that data." Of course it doesn't. The data isn't in ClickUp. It's in HubSpot, Salesforce, GA4, Google Ads, Meta, the data warehouse. Wave 2 tools don't connect to those. They connect to the agency's project board.
The structural mismatch nobody named
Both waves shared a structural mistake: they treated "client" as a user persona of "agency software" instead of treating "client" as a buyer who needs their own software.
The agency owner is a different user than the client. They have different jobs, different data sources, different questions, different success metrics. Building one product for both is like building one piece of software for both the seller and the buyer of a house. The seller wants a transaction tool. The buyer wants a decision tool. The same software cannot be both.
What both waves built, in the end, was good seller software with a "share" button. The buyer logged in, saw the seller's view, and left.
"ClickUp Brain answers questions of the agency's workspace. Next Best Action answers questions of your business. Same shape, opposite seat."
Wave 3: the client seat
The "what does the buyer need" generation.
Wave 3 starts with a different question: what does the marketing leader actually open every day? What signals would tell her the agency is working? What questions would she ask if she could ask anything? What approvals are sitting on her plate that she hasn't gotten to?
The product surface answers are different. Status of her business, not status of agency tasks. Ask any question of her data connected from her own systems. Approve work the agency proposes, with named humans behind every recommendation.
The Wave 3 surface looks like a marketing leader's home screen, not an agency's project board. The data sources include HubSpot, Salesforce, GA4, the warehouse, plus the agency's playbook. The questions are framed in the buyer's language. The strategist is in the loop on every Approve card and every high-stakes Ask answer, but the strategist is the supplier, not the user.
This is what we built in Next Best Action. It's also what Suits.ai is doing for competitive intelligence (the wedge: $8K per month replacing $25-30K hourly billings). What Code and Theory is doing for Qualcomm with The Machine. What 2X is productizing for the rollup model. The shape of the answer is consistent across the category. The execution differs by buyer.
What this means for the next decade
Wave 1 and Wave 2 portals will not disappear. They'll become a layer underneath, the way email became a layer underneath messaging. Agencies will keep using ClickUp or Asana to run their internal work, the same way they kept using email after Slack shipped.
What changes is what the client opens on Monday morning. The seat that's been empty for seven years gets filled. The product the client uses is purpose-built for their seat, not adapted from the agency's seat.
For agency operators, this is a category opportunity that's been sitting unclaimed for years. The agency that builds the client-facing surface gets a software-wrapped service business. The agency that doesn't, ships decks. Software-wrapped service businesses get acquired at 3-8x revenue. Pure services get 0.8-1.5x. Same retainer line item. Different multiple. We wrote about that math here.
For marketing leaders, the practical implication is to stop accepting "we'll set you up with a portal" as the answer to "how do I see what my agency is doing." That answer is Wave 1. The right answer is "you should have a surface that opens on your data, with my strategist in the loop, where you can decide instead of just observe."
The portals weren't bad. They were built for a seat that already had software. The seat that hasn't had software is the one that pays the bill.