# Chargeback defense for stores
Guide 9: win back the money stores lose to card disputes
https://www.everestchris.com/guides/chargebacks

## prompt 01 · step 2 · Connect Jev

connect to jev through my vercel ai gateway key (model typesafe-ai/jev). make up 3 disputes: one "item not received" where tracking shows delivery to the billing address, one "fraud" where the billing address and shipping address don't match and there's no cvv match, and one "not as described" where the customer emailed the store happy with the product a week after delivery. ask jev for each: fight, accept, or gather more. show me each answer and its probability.

## prompt 02 · step 3 · Write the rulebook

here is my rulebook for chargebacks, written in plain words: [paste your rulebook]

turn it into questions for jev (model typesafe-ai/jev through my vercel ai gateway key). for each dispute jev should answer: fight, accept, or gather more; win likelihood (low, medium, high); is the evidence complete (yes/no); is the deadline at risk (yes/no).

decision rules: prepare a fight only when jev is above 0.8 and win likelihood is at least medium. everything else goes to my review list with jev's answers.

save it as a skill called "chargeback rulebook" so i can make a copy per client.

## prompt 03 · step 4 · Write the response

for every dispute marked fight, write the response for the card network: one short paragraph with the order number, date, amount, product, and delivery details, then a numbered list of each piece of evidence with one line on what it shows. attach the evidence files from the order: [tracking, receipt, product page, customer messages]. only use evidence that exists right now. show me the response and attachments before anything is submitted.

## prompt 04 · step 5 · Find the stores

use my google places api key to pull every [furniture store / home goods store / boutique / jeweler / restaurant with online ordering] in [city]. save name, address, phone, website, rating, and review count. visit each website and note whether it has an online shop and what it runs on (shopify, square, woocommerce, other). filter to rating 4.0 or higher with a phone and website, put online shops first, and wait for my approval.

## prompt 05 · step 6 · Build the demo page

for [store name], make 3 mock disputes that fit a [store type]: realistic products and amounts, three different reason codes. for each, fill in which evidence the store has and doesn't have, and run it through my chargeback rulebook skill with jev. build a demo page: their name and logo from [website] at the top, each dispute with its reason code, the evidence checklist with ticks and crosses, jev's verdict and probability, and the full response for the strongest fight. put the total that would be won back at the bottom. mark it clearly as mock data, end with "send us your last 3 disputes and we'll run them for real, free." and my number: [your number]. host it on its own link.

## prompt 06 · step 7 · The postcard

design a 6x9 postcard per store. front: a screenshot of their demo page showing the 3 disputes and verdicts. back: "we set up chargeback fighting for [store name]. scan to see which disputes you'd win." with a qr to the demo page and my name and number: [your name], [your number]. show me 3 samples. after i approve, send the batch through my [lob / postgrid / stannp] account and log every send. also give me a one-line dm for each store's instagram with the same link.

## prompt 07 · step 8 · When they scan

message me whenever a chargeback demo page is opened, with the store name, their phone, and time on page. for stores that opened the page but didn't call after 3 days, draft one short follow up text from me for my approval. only one follow up per store.

## prompt 08 · step 10 · Fulfillment

[store name] connected their [stripe / shopify] account. every morning, check for new disputes. for each one, pull the order, payment details, tracking, and any customer messages, run it through their copy of my chargeback rulebook with jev, and write the response for every fight. put everything in front of me at least [n] days before each deadline. after i approve, submit the response through their [stripe / shopify] dashboard. at the end of each month, send me and the owner a statement of every dispute, the outcome, and the total won back.

## prompt 09 · The full prompt

you are my chargeback agent. run this pipeline and check in with me at the marked points.

niche: [furniture / home goods / boutiques / jewelers / restaurants with online ordering] city: [your city]

1. use my google places api key to pull every [niche] in [city]. save name, address, phone, website, rating, and whether the website has an online shop.

2. filter to: rating 4.0 or higher, phone and website listed, online shop first. show me the list and wait for my approval.

3. for each approved store, make 3 mock disputes that fit their store type, run them through my chargeback rulebook skill with jev (model typesafe-ai/jev through my vercel ai gateway key), and build a demo page: their name and logo, each dispute with its reason code, the evidence checklist, jev's verdict and probability, and the full response for the strongest one, marked clearly as mock data, ending with "send us your last 3 disputes and we'll run them for real, free."

4. design postcards: front is a screenshot of the demo page, back says "we set up chargeback fighting for [store name]. scan to see which disputes you'd win." plus a qr to the demo page and my number: [your number].

5. show me 3 sample demo pages and 3 sample postcards. after i approve, send the batch through my [lob / postgrid / stannp] account and log every send in a spreadsheet.

6. message me whenever a demo page is opened.

7. when a client connects their [stripe / shopify] account, check for new disputes every day, gather the evidence from their orders, run jev, and show me every response before it's submitted, at least [n] days before the deadline. keep a running total of what's won and send a monthly statement.

rules: never submit a dispute response without my approval. never list evidence the store doesn't actually have. no mail without my batch approval. cap api spend at [$x] without asking. save this workflow as a skill.