Which parts of a project does AI actually shrink?

Three of the six PROFIT Score numbers move when AI is real in your operating model: People (less team time), Investment (fewer agency and SaaS dollars), and Time (faster turnaround). Three numbers do not move: Return (the size of the prize), Odds (whether it works), and Fit (whether it advances strategy). AI changes how much a project costs and how long it takes. It does not change whether the project is the right thing to do.

How much does AI actually compress those three numbers?

Depends entirely on the kind of work. Repetitive analytical and content work (merchandising analysis, demand forecasting, attribution, landing pages, copy drafts) compresses hard — often 60% to 90% on team time. Hiring, market entry, brand pivots, and regulatory work barely compress at all because the bottleneck is human behavior, not analytical horsepower. The rest sits in the middle.

Does AI change my top priority project?

Usually no. The top of the stack mostly holds because Return, Odds, and Fit dominate the math, and AI does not change those. What changes is the middle of the stack. In the worked example here, a website redesign moves from rank 5 to rank 4, passing a new product launch — because AI compresses website work hard but barely compresses a product launch.

How often should I re-score with AI?

Quarterly. AI capability changes fast enough that a number that was a 4 on People last quarter might be a 2 this quarter as the team gets fluent with the tools. Re-scoring is roughly 90 minutes for 8 to 12 projects. Block the time on the same cadence as your strategic planning.

The Re-Rank: How AI Changes Your DTC Project Stack in 2026

Key Takeaways

AI changes three of the six PROFIT Score numbers: People, Investment, and Time. It does not change Return, Odds, or Fit.
Compression is uneven. Repetitive analytical and content work shrinks hard. Hiring, market entry, and brand pivots barely move.
The top of your stack mostly holds. The middle reshuffles. Quick wins get even better — strategic bets get a smaller boost.
In the example here, a website redesign jumps from rank 5 to rank 4, passing a new product launch. Same project, same outcome — different cost and time.
Re-score quarterly. Expect 1 or 2 extra priorities per quarter, not 5x throughput.

Last quarter you ranked your roadmap. Same business, same backlog, same strategy. But this quarter, your team is using AI seriously — not the demo-day version, the real thing. Claude in the daily workflow. AI on top of your data. AI drafting copy and analyzing spreadsheets. The merchandising team running demand scenarios in minutes instead of weeks. Marketing iterating on landing pages with a model instead of an agency.

Does the priority list change?

Yes. But not as much as you would think. And the changes are predictable.

Quick Recap

If you have not read the first post, start with The PROFIT Score. The short version: you score every project on six things — People, Return, Odds, Fit, Investment, Time — and multiply them. Higher is better. Sort and ship from the top. That is it.

The Thesis

AI changes three of the six numbers. It compresses People (less team time for the same work), Investment (fewer agency dollars and SaaS subscriptions), and Time (faster turnaround). It does not change Return (the size of the prize stays the same), Odds (whether the project actually works for your customers), or Fit (whether it advances your strategy).

So when you re-score the same backlog with AI assumed, the stack reshuffles — but it does not blow up. The most strategic projects stay near the top. Cost and time fall on roughly half the work. The middle of the list moves. We call that move The Re-Rank, and the right cadence to run it is quarterly.

Where AI Compresses Hard, Where It Does Not

The mistake most operators make is treating AI as a flat multiplier across the roadmap. It is not. Some work collapses; other work barely moves.

AI compresses hard. This is repetitive analytical and content work. Merchandising analysis. Demand forecasting. Attribution analyses. Drafting landing pages and comparison content. First-pass copy. Customer support replies. RFP scoring. Anything where the work is "look at data and produce a writeup" or "produce a variation of something we have shipped before." Expect 60% to 90% off team time, and a similar drop in agency cost.

The merchandising plan we walked through in Pillar #1 is the canonical example. Three people working a month each becomes one person working a week, with the model handling the data prep and first-pass scenarios.

AI compresses moderately. Vendor research. RFP scoring. Market scans. Competitive analysis. Internal tooling builds. Website redesigns. The analytical and content layers compress fast — but the human bits (negotiation, stakeholder alignment, QA, change management) do not. Net effect is roughly half-off, not 90%-off.

AI compresses minimally. Hiring. Market entry. Brand pivots. Retail expansion. Regulatory work. Anything that is bound by human relationships, consumer behavior, or external schedules. AI does not run the interview, does not change consumer preference, does not lobby the regulator. Expect 10% to 25% off team time at most, and almost no compression on calendar time.

If a founder tells you AI is going to 5x throughput across every project, ask which project. The answer reveals what they think AI is for.

Re-Score the Merchandising Plan

Take the same merchandising plan refresh from Pillar #1. Pre-AI it scored 1,512. What changes with AI assumed?

People. AI does the data prep, the SKU rationalization analysis, and the first-pass seasonal scenarios. Three people for a month becomes three people for a week. P drops from about 60 person-days to about 20.

Investment. A chunk of the freelance budget and most of the vendor analysis tooling becomes unnecessary. $250k drops to about $100k.

Time. Eight weeks of cycle time becomes four. Faster iteration, fewer meetings.

Return, Odds, and Fit. Do not change. Same project, same expected CM, same strategic position.

Re-multiply with the new bucket scores and the project goes from 1,512 to about 4,032. Roughly 2.7x higher than before. Not because the project got better — the project is the same. Because the cost of doing it dropped.

The Full Re-Rank

Now run the same exercise across all five projects from Pillar #1. Here is what happens to the order.

Rank	Pre-AI	With AI
1	Merchandising plan refresh	Merchandising plan refresh
2	Comparison landing pages	Comparison landing pages
3	Lower-COGS supplier change	Lower-COGS supplier change
4	Release new product line	Website redesign
5	Website redesign	Release new product line

The website redesign jumps from #5 to #4, passing the new product launch.

Why? Because AI compresses website work hard. The content, the design iteration, the code, the QA — all of it shrinks. Person-days drop from 800 to about 200. Agency cost drops from $1M to about $300k. Cycle time falls in half.

Meanwhile the new product launch barely moves. AI helps you write the brief faster and explore concepts faster. It does not change the supplier lead time, the retail buyer schedule, or whether consumers actually buy the product. The strategic value did not change, but the cost calculus around the redesign changed enough to overtake it.

That is The Re-Rank. Quiet, modest, predictable.

What This Means For You

Two takeaways for an operator running this in production.

You will ship more per quarter — but not 10x more. The realistic gain is one or two additional priorities per quarter that previously sat below your capacity line. That compounds. An extra growth lever per quarter for a year is meaningful. But it is not transformational, and any founder promising the board 5x throughput is setting up a hard Q3.

The order changes more than you would guess. Execution-heavy projects (websites, content systems, analytical builds) climb the list. People-and-judgment projects (hiring, retail expansion, brand pivots) stay roughly where they were. Re-score quarterly, because what was a 4 on People last quarter might be a 2 this quarter as the team gets fluent.

What AI Will Not Do

It will not compress hiring meaningfully. Interviews are still interviews. Reference calls still take a week to schedule. A VP search still runs four to six months end to end.
It will not change whether a new product wins with consumers. Demand validation takes cycles, and no amount of prompt engineering shortens that.
It will not fix bad strategy. If your Fit score was already a 3, AI does not bump it to a 7. The framework still requires you to know what you are trying to be.

Run Yours

The PROFIT Score works the same with or without AI in your operating model. You change three of the six numbers — People, Investment, Time — and re-rank.

Open the live PROFIT Score sheet. Run your top 8 to 12 projects through it twice — once at today's capability, once with realistic AI compression on the projects that actually warrant it. The reshuffling you see is the priority change worth acting on this quarter.

If your re-ranked list looks dramatically different from your current quarterly plan, that gap is the conversation worth having with your leadership team this week.

If you have not read it yet, the first post in this series is The PROFIT Score. Start there.

Frequently Asked Questions

Does AI ever change Return, Odds, or Fit?

Indirectly, on long horizons. Over multiple years, AI shifts which products are differentiated, which channels are competitive, and which capabilities matter — and that flows through to Fit. AI can occasionally raise Odds on data-rich projects where the bottleneck was analytical confidence. Return is mostly unchanged on a 12-month horizon. The first-order Re-Rank effect is on People, Investment, and Time. That is where 90% of the ranking change comes from.

How do I score AI-native projects (e.g., "build an AI agent for support")?

Same way as any other project. Six vectors, same anchors. The fact that the project uses AI in its delivery does not change the framework. What does change is that AI-native projects often have unusually low People-days, low Investment, and short Time, so they tend to score high — which is correct. They are a good use of capacity right now.

What if my AI capability is changing too fast for quarterly re-scoring?

In the first one or two quarters of serious adoption, monthly is fine. Once your team has stabilized on a stack and workflow, quarterly is enough. Re-scoring is cheap. Err toward more frequent if uncertain.

How does this affect agency relationships?

Materially. Projects that used to cost $80k to $100k in agency fees now cost $10k to $15k. Agencies that survive the next two years are the ones bringing strategic input AI cannot replicate. The ones selling person-hours have a problem. Worth raising in your next QBR with any agency partner spending more than $250k a year of your budget.