Analyst for Pricing Page Positioning

The pricing page is where positioning either pays rent or gets evicted. Three tier names, six feature rows, and one CTA per column have to do the work that fifteen sales calls do for an enterprise deal — and most founders write that copy in a Tuesday afternoon between a board prep and a hiring loop. An AI analyst won't fix that. But it will compress the loop from "ship something, watch conversion drop, blame the funnel" to "pressure-test the page before it goes live."

By "AI analyst" we mean a structured back-and-forth with a model that has your positioning brief, your competitor pages, and your ICP loaded as context — not a one-shot "rewrite my pricing page" prompt. The difference is the difference between a brainstorm and an audit.

The pricing page is positioning under maximum compression. Treat it that way.

What an analyst is good at, and what it isn't

Be honest about the boundary before you start. An analyst is excellent at consistency checks, alternative phrasings, and surfacing the assumptions buried in your tier names. It's bad at knowing what your buyers actually said in last week's calls, and it's worse at deciding which tradeoffs are commercially defensible.

If you ask the model to set the price, it will pick a round number and write a confident paragraph defending it. That paragraph is fiction. Pricing decisions come from win/loss data, willingness-to-pay research, and margin math — none of which the model has unless you give it.

The four-step loop

The order matters. Founders skip step two and go straight to step three, and end up with prettier copy that has the same underlying contradictions.

The prompts that actually work

Generic prompts produce generic output. The prompts below assume you've already loaded the context in step one — your brief, your current page, three competitor pages.

For the consistency check: "Read my pricing page against my positioning brief and homepage hero. List every contradiction, every term used differently across the three, and every claim on the homepage that the pricing page fails to substantiate. Don't suggest fixes yet. Just the list."

For tier names: "My current tiers are [X, Y, Z]. Generate five alternative naming sets where the tier names describe the buyer, not the feature count. For each set, name which buyer segment each tier targets and which one tier you'd cut if forced to ship two tiers."

For feature rows: "Here are my current feature rows for the [Tier] column. Rewrite each as: (a) the same row but in the buyer's language, (b) the same row but expressing the outcome instead of the capability, (c) the same row but as the smallest specific number you can defend. Don't editorialize."

For the objection stress-test: "You are a [ICP role] evaluating this pricing page after one demo. Write the three questions you'd ask in the next sales call that this page didn't answer. For each, point to the specific copy gap."

What the model gets wrong

Three failure modes show up reliably enough to plan for them.

Recurring analyst failures on pricing pages

The fix for all three is the same: review every output against the positioning you walked in with, not against the question of whether the new copy reads better. Better-reading copy that softens a defensible edge is a downgrade, not an upgrade.

C

The first draft the model gave me was beautifully written and described a company I didn't run. The fourth draft, after I kept feeding it our actual win/loss notes, was the one that started naming the buyer my sales calls actually closed.

What you still have to write yourself

The analyst doesn't replace three things, and pretending it does is how pricing pages drift into category-soup.

The first is the price itself. The number comes from your unit economics, your sales motion, and what your closed-won customers actually paid — not from the model's sense of what's "competitive."

The second is the lead-in line above the tier grid. That sentence is the most-read piece of copy on the page and it has to express your category position in your voice. Models default to "choose the plan that's right for you," which is wallpaper.

The third is the CTA verb. "Start free trial" vs. "Book a demo" vs. "Talk to sales" is a sales-motion decision dressed up as a copy decision. The model has no idea which motion you run.

The analyst saved us a week of tier-name iteration. It also tried to delete the one line on the page that was actually doing the differentiation work. Use it like a junior copywriter — fast, useful, needs supervision.

What to do this week

Block ninety minutes. Load the context — your brief, your current pricing page, three competitor pages — into a fresh chat. Run step two first; do not skip it. Take the consistency-check output to the founder or PMM who owns the brief and decide what's a real contradiction versus what's an artifact of how you described the company to the model.

Then run steps three and four. Treat every output as a draft to react against, not a replacement to ship.

The pricing page won't write itself, and an analyst won't write it for you. What it will do is shorten the time between "this copy is wrong" and "I know specifically which line is wrong and what the three better candidates are." For most founders, that's the bottleneck.