I'm closing out a quarter of writing about what's breaking in the product org with a piece about what's on the other side. This is the kind of post every thought leader writes at the end of a run, and most of them are vague enough that they can't be wrong. I'm going to try to avoid that.
Below are my specific predictions for what product organizations look like in the first half of 2028. Two years from the date this is published. Precise enough to be checked. Bold enough that I'll probably be at least partially wrong about them, which is fine, because being checkably wrong is more useful than being comfortably right.
Hold me to it. In July 2028, if this post is still on the site, I'll write the retrospective on what I got right and what I got wrong.
The short version
Ten predictions for what product organizations look like in the first half of 2028. Median team size drops from 12 to 6. Every product hire ships code in the first week. Evals become the highest-status, highest-paid skill in product. Per-outcome pricing is mainstream but hybrid. The roadmap becomes a public artifact, not a planning artifact. The PM/engineer/designer titles blur into Product Builder. Customer Success becomes outcome-attainment, not adoption. Quarterly OKRs survive only for board narrative. The Eval Engineer becomes the new Growth PM. The product org reports through the CEO directly more often than through a CPO sitting between. Each prediction has a specific test you can run in July 2028 to check whether I was right.
For the present-tense rebuilds these predictions roll forward, see The CPO Mandate 2026, The Eval-First Product Org, and Per-Outcome Pricing(coming May 18). For the role definition the predictions assume, see the PM Standard.
Prediction 1: The median product team is six people, not twelve
In 2022, the median high-performing product team was ten to twelve people: a lead PM, two to three additional PMs, a senior designer, one or two junior designers, a tech lead, four to six engineers, and sometimes a researcher. This was the shape of the "empowered product team" Cagan described, plus the organizational sediment most companies accumulate.
By 2028, the median will be six. A Product Builder lead, a second Product Builder or designer-Builder hybrid, three engineer-Builders, and a researcher who covers two or three teams.
The headcount reduction is not about layoffs. It's about the fact that six people with the 2028 tool stack ship more than twelve with the 2022 stack. The cost savings get reinvested in having more six-person teams, not in reducing total product headcount.
What the six-person team looks like in practice: every member can build prototypes. Every member can read and write evals. The team owns end-to-end outcomes, not feature areas. The team self-sufficiently ships from customer signal to production without handoff layers.
Test for this prediction: median product team size at top-quartile AI-native companies in July 2028.
Prediction 2: Every product hire ships code in their first week
By 2028, "ships code in the first week" will be a baseline expectation for any product role, not a notable skill. This will be true for PMs, designers, and researchers, not just engineers.
The hiring process will have a paid trial project as the primary evaluation. Whiteboard cases will have disappeared. Case interviews will be residual for a small number of traditional consulting-adjacent product roles.
Test for this prediction: in July 2028, pick ten top-tier product roles at public AI-native companies. Check whether the job descriptions require shipping code.
Prediction 3: Evals become the highest-status skill in product
The eval engineer (or eval product manager, however the title shakes out) will be the highest-paid, highest-status role in product by 2028. It will be what "growth PM" was in 2018 and what "AI product manager" briefly was in 2024.
The reason is straightforward: eval systems determine what quality means for a product, and in a world where the software does the work, quality is the product. The person who owns quality owns the most important thing.
Eval engineering will not be an engineering discipline. It will be a product discipline with engineering depth. The senior eval engineer will be paired with the senior PM the way design leads were paired with PMs in 2015.
Test for this prediction: check the compensation bands for "eval engineer" or equivalent titles at top-quartile AI companies in July 2028. Compare to compensation bands for senior PM, senior designer, and senior engineer at the same companies.
Prediction 4: Per-outcome pricing is mainstream, but hybrid
By 2028, most enterprise software above $50K ACV will have a per-outcome component. Not all of it will be pure per-outcome. Most will be hybrid: a platform fee plus per-outcome overage, with committed minimums for enterprise customers.
Pure SaaS pricing will still exist, especially for tools whose value is genuinely about augmenting human work rather than replacing it. But the dominant pricing shape for enterprise software that touches knowledge-work automation will be hybrid.
The shift will not be graceful. 2026 and 2027 will see pricing experimentation that fails in public. A few high-profile SaaS companies will try to shift to pure per-outcome, discover the forecasting nightmare, and retreat to hybrid. By 2028, most of the right answers will be known.
Test for this prediction: survey the top 100 enterprise software companies by revenue in July 2028 and check their pricing models. Hybrid should be the mode.
Prediction 5: The roadmap is a public artifact, not a planning artifact
Most product orgs by 2028 will publish a real-time public roadmap that shows what bets they are currently running, what bets retired, and what bets are in prototype-phase. This will be the default level of transparency for any company that sells to developers or technical buyers.
The roadmap will no longer be a commitment document. It will be a record of current belief. Customers will expect to see it updated weekly. Competitors will watch it constantly. The old "internal roadmap" and the old "stale marketing roadmap" will both be gone.
This is already happening at companies like Linear, Vercel, and Raycast. By 2028, it will be the norm for any product-led company with a technical audience.
Test for this prediction: in July 2028, check the top 50 product-led B2B companies for a public, real-time-updated roadmap. More than half should have one.
Prediction 6: The monolithic product function splits into three
Today's single "Product" function reporting to a CPO will have split into three related but distinct functions at most large companies.
- Product Engineering. The teams that actually build. Product Builders, engineers, designers. This is what most people think of today as "Product" plus a chunk of today's Engineering.
- Quality and Economics. The Quality Spine and Economics Unit I wrote about in the Eval-First Product Org post. Owns evals, unit economics, model routing, and quality dashboards.
- Discovery and Strategy. The smaller, strategic function that owns customer understanding, market positioning, and long-range bets. Closer to a research-plus-strategy function than traditional product management.
All three will often report to a CPO or a Chief Product and Technology Officer (CPTO), but they will have separate charters, separate teams, and separate success metrics. The blurring of these functions that characterizes most product orgs today will be seen as a legacy artifact.
Test for this prediction: in July 2028, look at the org structure of five to ten AI-native companies above $500M revenue. At least three should have a recognizably tripartite product function.
Prediction 7: The APM program is fully extinct at top companies
Every major APM program at a top-tier tech company will be gone by 2028. Google APM, Facebook RPM, the handful of others that still exist, will either be discontinued, rebranded into a Product Builder apprenticeship, or quietly wound down.
The companies that try to keep the classic APM curriculum will find that their graduates can't compete with the ones coming out of Product Builder apprenticeships for entry-level roles. This will be evident by early 2027 and fully visible by 2028.
Test for this prediction: in July 2028, check whether Google APM, LinkedIn's successor to their APM program, and two to three other major programs are still running the 2022 curriculum or have visibly rebuilt.
Prediction 8: The CPO role has meaningfully changed hiring criteria
By 2028, CPO hiring will have shifted. The current archetype (a senior product leader who ran a large org at a previous company) will still exist, but it will be one of several. Two new archetypes will be mainstream.
- The builder-CPO. A senior Product Builder who's scaled from building individual prototypes to running a small-but-technical product org. Often coming from an engineering-adjacent background. Hiring criteria emphasizes shipping, not managing.
- The operator-CPO. A leader with heavy operations and services experience, specifically for Service-as-Software companies where running a production system is the core challenge. Hiring criteria emphasizes quality, cost, and delivery at scale.
Companies will consciously choose between these archetypes based on their stage and business model. The default "hire a VP of Product from a more successful company" pattern will be seen as insufficient for AI-native businesses.
Test for this prediction: in July 2028, look at the last 20 CPO hires at AI-native companies above $100M revenue. The traditional archetype should be fewer than half.
Prediction 9: Product operations as a standalone function is mostly gone
Product Ops as a named team will have disappeared from most mid-sized and smaller companies by 2028, absorbed either into the Quality and Economics function, into engineering ops, or into centralized finance partnerships.
At large companies, a small Product Ops team may persist, but its charter will be much narrower: tooling and infrastructure for the product function, not process management.
Test for this prediction: in July 2028, check the org charts of 20 mid-sized product organizations (between $50M and $500M revenue). Product Ops as a named team should be present at fewer than 20% of them.
Prediction 10: One major incumbent SaaS company will have failed the transition
By mid-2028, at least one well-known mid-to-large SaaS company ($500M+ ARR) will have publicly struggled with the transition to Service-as-Software. This will not be a full bankruptcy, but it will be a visible failure: missed earnings, executive churn, an activist investor, or a down round.
The failure will be attributed in the press to a "pricing transition gone wrong" or a "product strategy misstep," but the underlying cause will be that the company tried to bolt outcome-based pricing onto a SaaS architecture without the two-horizon rebuild I wrote about in the Service-as-Software field guide post.
I'm not going to name which company I think is most at risk. I have guesses. They would be unkind to publish. The prediction is directional, not personal.
Test for this prediction: in July 2028, check whether any major public SaaS company has had a visible stumble that press coverage attributes to pricing or product model transition.
The predictions I'm most worried about
Two of the predictions above are the ones I'm least confident in, and I want to flag them.
Prediction 2 (every product hire ships code in their first week) might be too aggressive. The tooling for PMs to ship code is improving rapidly, but the cultural adjustment at many companies is slower than the tool curve. The more likely 2028 state is that shipping code is a strong expectation for senior PMs and optional for junior ones, with the transition still incomplete.
Prediction 6 (product function splits into three) might be too structural. The underlying trend is right (specialization around evals and economics is happening), but the exact org structure that emerges might be different from the tripartite split I described. Could easily be two functions or four, or different groupings.
If either of these is wrong in July 2028, I won't be surprised.
What I'm most confident in
The things I'd bet real money on:
- Product teams get smaller and flatter (prediction 1).
- Evals are the highest-status skill in product (prediction 3).
- Hybrid pricing is mainstream (prediction 4).
- APM programs as currently structured are gone (prediction 7).
These four are either already happening at the leading edge or are so clearly forced by the underlying economics that I have trouble imagining them not playing out. Happy to be wrong. Don't think I am.
What this means for you right now
If you're running product today and you want to be ready for 2028, the work is not waiting until 2027 to start adapting. The work is this year.
- Hire one eval lead. Give them senior status. Publish their scorecard.
- Run one Product Builder apprenticeship. Start with two apprentices.
- Model one of your products at per-outcome pricing. See what it reveals.
- Cut your next product team's headcount target by 30%. See what happens.
- Publish your real roadmap. Not the marketing one. The actual one.
Doing one of these this quarter puts you ahead. Doing three of them by year-end puts you in a different league than peers who are still running the 2022 playbook.
The companies that make these moves in 2026 will have functional 2028 product orgs in 2027. The ones that wait will be re-orged by their boards in 2027 anyway, but they'll be behind, and the shape of the re-org will be imposed on them rather than chosen.
Pick the choice. Make the moves. See you in 2028 with the retrospective.
This is the last post in the Service-as-Software series for now. The full toolkit, including every template referenced across the series, is at falkster.com/toolkit. Thanks for reading.
Further reading
Frequently asked
What are the ten predictions for product orgs in 2028?+
Median product team size drops from 12 to 6. Every product hire ships code in their first week. Evals become the highest-status, highest-paid skill in product. Per-outcome pricing is mainstream but hybrid (platform fee plus overage). The roadmap becomes a public artifact, not a planning artifact. The PM/engineer/designer titles blur into Product Builder. Customer Success becomes outcome-attainment, not adoption. Quarterly OKRs survive only for board narrative. The Eval Engineer becomes the new Growth PM. The product org reports through the CEO directly, not through a CPO sitting between.
Why will the median product team shrink from 12 to 6 by 2028?+
Because six people with the 2028 tool stack ship more than twelve with the 2022 stack. The reduction is not about layoffs, the cost savings get reinvested in having more six-person teams, not fewer total product headcount. What the six look like: every member can build prototypes, every member can read and write evals, the team owns end-to-end outcomes, no handoff layers.
Will every PM really need to ship code by 2028?+
Yes. 'Ships code in the first week' will be a baseline expectation for any product role, not a notable skill. This will be true for PMs, designers, and researchers, not just engineers. Hiring processes will use a paid trial project as the primary evaluation. Whiteboard cases will have disappeared. The test in July 2028: pick ten top-tier product roles at public AI-native companies and check whether the JDs require shipping code.
Why will the Eval Engineer be the highest-status PM role by 2028?+
Because eval systems determine what quality means for a product, and in a world where the software does the work, quality IS the product. The person who owns quality owns the most important thing. Eval engineering will not be an engineering discipline; it will be a product discipline with engineering depth. Senior eval engineers will be paired with senior PMs the way design leads were paired with PMs in 2015.
Will pure SaaS pricing survive in 2028?+
Only for tools whose value is genuinely augmenting human work rather than replacing it. The dominant pricing shape for enterprise software that touches knowledge-work automation will be hybrid: platform fee plus per-outcome overage with committed minimums for enterprise. Pure per-outcome will exist but most companies that try it will retreat to hybrid after discovering the forecasting nightmare.
Why will the roadmap become a public artifact?+
Because the roadmap as a private planning document is increasingly indefensible. AI-native companies competing for trust with customers will ship public roadmaps as differentiation. The internal roadmap shrinks to the bets-with-active-decision-deadlines layer (the outcome ledger). The strategic narrative goes external. The board still gets a quarterly story but it's downstream of what customers can already see.
How can I check these predictions in July 2028?+
Each prediction includes its own check. Median team size at top-quartile AI-native companies. JD requirements for top-tier product roles. Compensation bands for 'eval engineer' titles. Pricing model survey of the top 100 enterprise software companies. The post is timestamped, if it's still on the site in July 2028, the retrospective will be too.