A product list becomes useful when it removes ambiguity.
A product catalog for recurring purchases should tell an assistant what is acceptable to buy again. Names alone are rarely enough. Coffee may mean beans or capsules; printer paper may mean a ream or a case. The goal is to keep those decisions from being made differently on every run.
Start with the products you already know. A saved link, a product card or a row imported from a CSV can be an entry point, but the final selection still needs an exact product and variant. Importing a spreadsheet does not establish that a listing is current, deliverable or authorized for purchase.
Keep the catalog small at first. Five well-specified essentials create a more reliable routine than fifty vague entries. Expand after the first review shows that the product cards, quantities and buying rules represent what you intended.
Use fields that answer real buying questions.
This is an example planning structure, not a promise that every column is automatically recognized by an importer. Map the information to the fields and controls available in your account. Keep credentials, payment-card information and unrelated customer data out of the file.
Add a short reason for any strict requirement. For example, a replacement filter must fit a particular device. The assistant should treat that as a constraint, even if another product has a better price or rating.
| Field | Illustrative value | Why it matters |
|---|---|---|
| Product and variant | Coffee beans, medium roast, 12 oz | Prevents a format or size substitution |
| Product URL | The exact retailer listing | Gives the assistant a concrete starting point |
| Quantity | 2 bags | Separates item count from package size |
| Substitution rule | Ask before changing brand | Defines the fallback |
| Review threshold | $30 delivered for this line | Makes a price change visible |
| Category | Office kitchen | Helps later expense review |
Choose a closed catalog or permission to explore.
A closed selection means the bot should stay within the products you selected. If none are available, the useful outcome is a clear exception for you to review. Buying a different item just to complete the run defeats the purpose of the catalog.
A flexible selection allows alternatives within your preferences. Be precise about what can change: seller, brand, pack size or product type. Permission to find another seller should not silently become permission to change the product.
For a mixed sheet, you may need separate routines. Keep compatibility-sensitive items in a strict routine and flexible household supplies in another. That makes the intended behavior easier to understand and review than a single long paragraph with exceptions.
Keep quantity, frequency and budget separate.
Two boxes each month describes a quantity and a schedule. It does not tell the bot what to do if a box doubles in price or if you have three unopened boxes left. Decide how you will check stock and how you will pause the next request.
Use a line-item ceiling when one product must not exceed a certain amount, and an overall order ceiling for the basket. Make sure the approval screen's treatment of delivery and fees matches your expectation. A catalog entry is a preference; the actual repeat authorization defines the permission to spend.
As an example, a small office could keep coffee and paper in one weekly review, while replacement hardware remains manual. The office does not need every product on the same schedule simply because they share a catalog.
Build one catalog entry all the way through.
Take a product you already buy, such as medium-roast whole-bean coffee in a 12-ounce bag. Begin with the exact listing rather than a category search. Save the product name, the format, the package size and the link. If the retailer offers several variants on one page, select the actual variant rather than assuming the link alone preserves it.
Next, define the unit you intend to buy. Two bags is a quantity of two packages, not two ounces and not two cases. Check how the product card represents that choice. If the listing is a multipack, record how many bags arrive in one purchased unit so the repeat quantity remains understandable.
Add the acceptable fallback. Perhaps another seller of the same product is fine, but a different roast needs review. That rule is much more useful than alternatives allowed without qualification. The assistant needs to know which characteristics define equivalence for you.
Finally, put the entry into a trial basket and inspect the resulting quote. The catalog is doing its job only if the card, quantity and price correspond to the thing you intended to buy. Resolve any ambiguity in the saved entry before building a large sheet around the same pattern.
Prepare a CSV as source material, not as an authorization.
If you already have a spreadsheet, begin by removing duplicate and obsolete rows. Keep one row per distinct product variant. Separate product names, URLs and quantities into consistent columns rather than combining the whole instruction into a single cell. This makes the information easier to inspect even when the importer requires manual mapping.
Preserve identifiers that distinguish products. A model number, a specific size or a compatible device can matter more than a marketing name. For a product with several color or pack-size options, make those distinctions explicit. A visual card can then help you recognize the entry without hiding the underlying detail.
Treat prices in the sheet as historical references unless the workflow explicitly refreshes them. A saved amount does not prove today's availability or payable total. Likewise, a valid URL does not establish that the retailer will deliver to the intended destination. Those checks belong to the actual buying step.
Review the imported or selected products before attaching the sheet to a recurring bot. Importing data should not silently become permission to spend. Keep the catalog, the schedule and the repeat authorization as separate decisions, even if the interface lets you configure them in one flow.
Write substitution rules at the right level.
There are several kinds of substitution, and they are not equally safe for every product. Changing the seller while preserving the exact item differs from changing the package size, which differs again from choosing another brand. A broad instruction to find something similar leaves too much of that decision implicit.
For consumables, a larger package might be acceptable if the product is identical, storage is available and the total remains within the approved boundary. For replacement parts, even a visually similar product can be unsuitable. For personal preferences, such as a particular flavor, another brand may be technically equivalent but still unwanted.
Write the narrowest rule that solves your problem. For example: same product from another seller is acceptable; changing size requires review; changing flavor is not acceptable. Map those preferences to the controls actually supported by the product. If a rule cannot be represented reliably, keep the purchase under manual review.
Do not assume a spending ceiling makes substitutions harmless. The wrong replacement part can be inexpensive and still useless. Product identity, quantity and cost each need their own check. That is why a good catalog captures what makes an item acceptable, not merely how much you are willing to spend.
Keep the catalog useful after the first month.
A catalog needs occasional maintenance because products and requirements change. A retailer may replace a listing, a manufacturer may change package size or your office may switch equipment. When a repeat order fails or looks different, check the source entry instead of repeatedly patching the individual run.
Assign a clear owner if more than one person uses the catalog. The owner does not have to approve every order, but someone should know which entry is current and whether an old product should be removed. Multiple nearly identical entries make it easy to select an outdated variant.
Use descriptive product names that distinguish similar items without turning each row into a paragraph. Keep longer reasoning in the relevant notes or routine instructions where supported. The visual card should make selection quick, while the underlying data makes the selection precise.
When should a product leave the repeat catalog? When you no longer use it, its compatibility is uncertain or each purchase requires a fresh decision. Move that item back to a reviewed one-off request. A smaller catalog of dependable favorites is more useful than a comprehensive archive that the assistant must reinterpret every time.
Review the first basket as a catalog test.
Check the product image against the exact text and variant. Images are useful for recognition but can show a different pack or a generic family of products. Verify the quantity in the order rather than counting objects in the picture.
If something is wrong, correct the underlying saved product or rule before approving the next repeat. A one-off correction in chat may solve today's basket without fixing tomorrow's routine. Revisit saved links when a retailer replaces a listing or a manufacturer changes the package.
Jiclo connects saved products and product sheets to shopping routines, with review and explicit permission as separate steps. Use the run history and order record to check what was requested and what actually happened. For current access, supported behavior and costs, consult the services and pricing pages.
