A decoy is an extra option that almost nobody buys. You put it next to the offer you want people to pick, and that offer starts to look like the obvious, smart choice. My Claude prompt designs one for your business, with your own prices.
Locked. The prompt, and the link that loads it in Claude, open with your email.
Free with your email. Read it first: 3 real answers from Claude and the rules that keep a decoy honest are below.
On your phone? Copy the prompt and paste it into the Claude app. It works there too.
The Economist test100 MIT students, the same prices. Flip the decoy:
Web only $59
16%68%
Print only $125decoy
0%gone
Print + web $125
84%32%
Revenue from 100 students$11,444$8,012
Nobody picked print only. It was there to make print + web look like a free upgrade.Same prices, decoy removed. The option nobody picked was worth +43% in revenue.
Dan Ariely, Predictably Irrational (2008)
In one sentence
What a decoy is
A decoy is an option you add so that the one you want people to pick looks like the obvious choice.
Almost nobody buys it, and that's fine. Its job is to change what your best seller gets compared to. There are two kinds, and the prompt picks the one that fits your business.
The show-off
A far bigger, real version of your offer, priced about 3 to 10 times your best seller. Next to it, your best seller feels reasonable. It works as an anchor: a big number first makes the next one feel smaller.
Example: Tesla's Roadster Rocket, invite only, no public price, next to the Performance at $50,000 to reserve.
An option at about the same price as your best seller, but clearly worse. Next to it, your best seller looks like a free upgrade. This is the decoy effect researchers first measured in 1982.
Example: The Economist's print only at $125, next to print + web at $125.
Tesla's Roadster: three versions, one that just drives
On 15 Oct 2026 Tesla showed the new Roadster in three versions. The Performance is road ready. The other two add thrusters, and the wildest one has no public price.
The show-off
Rocket
1 seat, thrusters all around the car to lift it and turn it
0 to 60 mph in under 0.59 s, top speed over 230 mph
Not road legal
Invite onlyNo public price
Limited release
Thrust
2 seats, cold-gas thrusters at the back
0 to 60 mph in under 0.9 s
$250,000just to reserve
The one that drives
Performance
2 seats, road ready, active aero
0 to 60 mph in under 1.6 s
$50,000to reserve: $5,000 now, the rest within 10 days
The Rocket has two jobs
1
Attention
Over 25 million views on Tesla's YouTube channel alone, across 7 Roadster videos (my count on 20 Oct 2026). And the headlines were all about the thrusters.
2
The anchor
Next to a car you can only get with an invitation, $50,000 to reserve the one that drives starts to sound reasonable.
Tesla hasn't called the Rocket a decoy: this is my reading of the menu. The amounts are reservations. No final price was announced for any of the three versions, and production is planned for 2028, according to Musk. Facts as reported by Electrek, CNBC (its headline calls it a “flying car”), stock3 and Híbridos y Eléctricos. Views counted on Tesla's YouTube channel.
The near-twin
The Economist test: the option nobody picked
In his book Predictably Irrational (2008), Dan Ariely showed 100 MIT students a real offer from The Economist. Two of the three options had the same price.
Web only$59With the decoy: 16% Without it: 68%
Print only decoy$125With the decoy: 0% Without it: removed
Print + web$125With the decoy: 84% Without it: 32%
84%chose print + web when print only was on the menu
32%chose it once the option nobody picked was removed
+43%revenue with the decoy: $11,444 against $8,012, from the same 100 students
Where the idea comes from. Huber, Payne and Puto first measured it in 1982, in the Journal of Consumer Research (“Adding Asymmetrically Dominated Alternatives”). They called it asymmetric dominance: the decoy is clearly worse than the option you want picked, so picking that option feels like an easy win.
✦ Free · the prompt
Make your best seller the obvious choice
🦆🏷️🏆📈
The decoy prompt: fill in four lines with your real prices.
How to run it in Claude on the web, the app or Claude Code.
One link that loads it in Claude Code.
You're also joining AI Frontier, my weekly email: one email every Thursday, unsubscribe anytime. Privacy
The prompt
Copy it, fill in four lines
Replace the four lines in brackets with your business, or leave them as they are and Claude will ask you. Use your real prices: the prompt works with your numbers. Everything else stays exactly as written.
Locked. The decoy prompt, ready to copy, opens with your email. You fill in four lines with your prices.
I ran the prompt, word for word, on three invented businesses. Below are Claude's real answers, shortened: every name, price, badge and decoy is exactly what Claude wrote. Read the honesty checks too: that's where Claude pushes back.
Wedding photography studio
Claude picked the show-offInvented business
Sells
Wedding photography, a two-person studio near Denver, about 30 weddings a year
Best seller
Full Day, $4,800: 8 hours, two photographers, around 600 photos
Also sells
Essentials, $2,400 (4 hours, one photographer). Album add-on, $900
Buyers
Couples aged 28 to 38 who compare 3 or 4 photographers on Instagram
Claude's diagnosis
Full Day is compared only to Essentials: exactly 2x the price for 2x the hours. With no bigger option above it, $4,800 is the ceiling of the range.
Why the show-off
A near-twin could look cynical if a couple ever notices, which is bad for a business that lives on referrals. A real, bigger option makes Full Day the grounded middle pick.
Essentials$2,400
A clean, honest record of your day.
4 hours, 1 photographer, about 250 photos
Most Booked
Full Day$4,800
Full coverage, two photographers, nothing missed.
8 hours, 2 photographers, about 600 photos, delivered in 6 weeks
the decoy
The Complete Story$15,900
For couples who want their whole weekend documented.
2 days (rehearsal + wedding), 2 photographers, about 1,000 photos, 2 albums
Honesty check. Use “Most Booked” only if it's true, otherwise “Our Favorite” or “Recommended”. Publish The Complete Story only if your second shooter can cover two-day weekends.
The test. At about 30 weddings a year, 14 days catch only 1 or 2 bookings. So: track quote requests for the 14 days, and judge revenue over 60 to 90 days.
B2B web agency
Claude picked the show-offInvented business
Sells
Webflow websites for B2B software companies, a team of 6
Best seller
Growth Site, $32,000: up to 15 pages, copy included, 8 weeks
Also sells
Launch Site, $14,000 (up to 6 pages, the client writes the copy). Care plan, $1,800 a month
Buyers
Founders and heads of marketing at SaaS companies after a funding round, comparing 3 agencies
Claude's diagnosis
Next to Launch Site alone, buyers read $32K as “the expensive option” and the top of your range: the riskiest frame for a $32K decision.
Why the show-off
Launch Site is honestly cheaper and honestly less, so a near-twin doesn't fit. A real, bigger offer above Growth Site makes $32K look like the smart middle.
the decoy
Enterprise Site$110,000
For companies expanding into multiple markets or product lines.
Up to 60 pages, 3 sites, full copywriting and localization, dedicated strategist
Recommended for Series A-C teams
Growth Site$32,000
Everything you need to launch and scale content without rebuilding again.
Up to 15 pages, copywriting included, CMS for blog and case studies, 8 weeks
Launch Site$14,000
A fast, lean site when you're writing the copy yourself.
Up to 6 pages, you supply copy, 4 weeks
Honesty check. Offer the $110K project only if your team of 6 would really take it; if not, lower scope and price. The badge says “Recommended for Series A-C teams” instead of “Most Popular”, because it's true.
The test. Keep it if Growth Site's share of wins and the revenue per proposal go up. If the overall win rate drops sharply, bring Enterprise down toward 2.5 to 3x.
Neighborhood yoga studio
Claude picked the near-twinInvented business
Sells
Yoga classes, 32 a week, 20 mats per class
Best seller
Unlimited membership, $139 a month
Also sells
Drop-in class $24, 10-class pack $200, intro offer of 2 weeks for $39
Buyers
Working adults 25 to 45 who live or work within 15 minutes
Claude's diagnosis
The 10-class pack feels like the safe, bounded choice ($20 a class), while $139 a month feels open-ended. So intro students often pick the pack.
Why the near-twin
A $1,000+ “VIP unlimited” tier would be a bluff at a 20-person studio. A pass priced just under $139 that's clearly worse for regulars makes Unlimited look like a free upgrade.
10-Class Pack$200
Low-commitment option for occasional practice.
10 classes, any level, valid 3 months
the decoy
Monthly 8-Class Pass$136a month
For a steady 2x/week habit.
8 classes a month, resets monthly, no rollover
Best Value for Regulars
Unlimited Membership$139a month
For just $3 more than the 8-Class Pass, come as often as you like.
Unlimited classes, all 32 a week, cancel anytime
Honesty check. Check that $17 a class ($136 for 8 classes) still covers instructor pay. And no “Most Popular” badge until sales data backs it.
The test. If more than about 15 to 20% of new members pick the 8-Class Pass, it's eating memberships: move its price closer to $139.
Run on 20 Oct 2026 in Claude Code 2.1.280 with Claude Sonnet 5, in a clean setup: no tools, no files, nothing but the prompt. The businesses and their numbers are invented for the examples. Each answer came in one go, with no questions back.
The rules
Honest decoys only
The prompt checks these on every answer and flags what breaks them. In the examples above, Claude flagged badges you can't back up yet, and an offer a small team might not deliver.
✓
It's realIf someone buys the decoy, you deliver it, gladly. If you wouldn't sell it, it can't be on the menu.
✓
No fake scarcityNo “last few spots” unless it's true.
✓
No hidden feesThe price on the menu is the price.
✓
No false badges“Most popular” only when your sales say so. Until then, “Recommended” is the honest word.
✓
Keep it only if it worksMeasure the share of buyers who pick your best seller, average order value and revenue per visitor.
FAQ
Short answers
Is this the decoy effect from psychology?
The near-twin is: Huber, Payne and Puto described it in 1982 and called it asymmetric dominance. The show-off works more like an anchor: a big number first makes the next one feel smaller.
Is the Economist test solid?
It's a classroom test from Dan Ariely's book (2008): 100 MIT students and a real Economist offer. Some of Ariely's other studies have been questioned since 2021. I quote this one from the book, as an illustration. Your own test is what counts.
Isn't this manipulation?
Only with a fake decoy. A real option you'd gladly sell gives buyers something honest to compare against. The prompt flags fake scarcity, hidden fees and false badges.
What if nobody buys the decoy?
That's normal. In the Economist test, nobody picked print only. Watch the share of buyers who pick your best seller: that's the number the decoy should move.
Do I need Claude Code?
No. It's a plain prompt: claude.ai, the Claude app and Claude Code all work.
Does it work for a proposal?
Yes. Step 4 writes the words for a pricing page or a proposal, and the menu works the same way in a quote.