Your product page ends with a handmade “You may also like” row that has not changed since last spring. Cart shows two static upsells a marketer picked in a meeting. Thank-you page pitches a product the buyer already owns. That is not personalization. That is a poster.
Real recommendations follow what someone browsed and what people like them bought. Generatify’s WooCommerce product recommendations are built from that behaviour, placed on product, cart, and thank-you pages, and fall back to best sellers when a visitor is brand new.
Generatify is the DaftPlug AI plugin. This post is about AI product recommendations: browse and order history, three placements (product, cart, thank-you), and cold-start best sellers. Not the page agent, not the helpdesk agent, not draft-in-editor, not translate, not bulk alt text, not the usage cost dashboard, not comment toxicity. For an adjacent “acts on the shop” Generatify feature, keep the WordPress page agent that navigates filters and cart steps nearby. That agent helps someone move. This feature suggests what to look at next.
What Generatify actually ships here
Stay inside the real feature set from the product page:
- Recommendations built from real browsing and order history
- Placements on product, cart, and post-purchase (thank-you) pages
- Cold-start fallback to best sellers when there is not enough personal signal yet
- Part of Generatify’s WooCommerce-aware AI toolkit (you connect your own model key on Pro; you pay the provider directly)
Do not invent affinity algorithms, email flows, or “net lift dashboards” the product page does not claim. The promise is behaviour-based picks in the three shop moments that matter, with a sensible fallback.
Why behaviour beats a static upsell plugin
Static lists rot. Seasons change. Inventory changes. A fixed related-products widget becomes folklore.
Browse history is intent you already captured. Someone who viewed trail shoes and socks told you more than a category checkbox on the product editor ever will.
Order history is proof. Post-purchase and cart moments are where “next logical product” earns money without feeling like a pop-up lottery.
Cold start is real. First-time visitors have no history. Best sellers beat empty shelves or random catalog noise.
You already pay for AI elsewhere in Generatify. Recommendations sit beside assistant, helpdesk, and editorial tools under one membership instead of bolting on a separate personalization SaaS with another renewing seat.
It stays on the storefront. Product, cart, and thank-you are shopper-facing surfaces. You are not asking the merchant to manage a fifth dashboard to move three widgets.
If you sell through multiple catalogs or switch themes often, treat recommendations as part of the storefront QA checklist after every theme update. A new product template that drops hooks will look like an AI failure when it is really a template failure.
How to set up recommendations in wp-admin
Assume WooCommerce is running and Generatify is installed with a Pro model connector configured (your own ChatGPT, Claude, Gemini, or Kimi key per product FAQ).
1. Confirm the shop surfaces exist
Before toggling AI, confirm:
- Product pages render normally.
- Cart and checkout are reachable.
- Thank-you / order-received templates are not replaced by a page builder blank that strips Woo hooks.
If a custom thank-you template removed WooCommerce hooks, fix that first. Recommendations cannot appear on a template that never loads the placement.
2. Open Generatify product recommendations
In wp-admin, open Generatify and find the product recommendations feature (AI product recommendations on the feature list). Enable it for the store. Keep the mental model simple: behaviour in, placements out, best sellers when behaviour is missing.
3. Enable the three placements on purpose
Turn on only what you will actually review:
- Product page for “next relevant item” while intent is hot.
- Cart for complementary adds before payment anxiety spikes.
- Thank-you for a post-purchase next step that does not undo the order they just placed (think care products, accessories, replenishment, not a random unrelated hero SKU).
Disable a placement if your theme already overcrowds that template. Empty space with one strong row beats three competing upsell plugins.
4. Respect cold start
On a fresh session with no browse or order signal, expect best sellers. That is a feature, not a bug. Check that your best-seller base is sane (not a discontinued SKU, not an internal test product).
As browse and order history accumulate for a shopper, recommendations should reflect that history rather than the static poster row you used to maintain by hand.
5. Sanity-check with three shopper scripts
Run these on a private window and a returning-customer window:
- Cold visitor: land on Home → open a product → confirm a best-seller-style row appears rather than blank.
- Browser: view two products in a niche (for example, two camera bags) → open a third related product page → confirm suggestions feel closer to that niche than to sitewide bestsellers alone.
- Buyer path: add an item → open cart → complete a test order → confirm thank-you suggestions are present and not nonsense relative to what was bought.
You are not trying to prove a research paper. You are trying to catch “recommends the same product already in cart” and “recommends out-of-stock junk” early.
6. Keep adjacent Generatify tools out of this pass
Do not also rebuild helpdesk macros, page-agent limits, bulk alt text, and translation memory in the same afternoon. Ship recommendations. Watch a week of real carts. Then decide whether the page agent or helpdesk needs attention.
A useful extra check: open the same product on mobile and desktop. If the recommendation row collapses under a crowded related-products widget from the theme, hide or simplify the theme widget so Generatify owns that surface.
What good looks like
- Product pages show a recommendation row driven by browse/order behaviour when signal exists.
- Cart shows complementary suggestions without blocking checkout.
- Thank-you shows a post-purchase next step instead of a dead end.
- Brand-new visitors still see best sellers instead of an empty module.
- Out-of-stock and draft products do not dominate the row (fix catalog hygiene if they do).
- You are not maintaining a seasonal “related products” spreadsheet by hand.
- Model usage stays on your provider key with no Generatify markup on tokens (per product positioning).
If every page shows the same five SKUs forever, you are either always cold-starting or your history signal is not reaching the feature. Recheck that browse events can fire on the theme you use (aggressive optimization or broken Woo templates can starve signals).
Pitfalls (the ones that bite)
Custom templates that strip Woo hooks. Especially thank-you. Fix the template.
Expecting miracles on day one with no traffic. Cold start means best sellers until history exists.
Stacking three upsell plugins on cart. Pick one owner for recommendations.
Recommending the product already in the cart as the hero suggestion. If you see that, adjust placement rules or catalog relationships and retest.
Using this as a substitute for merchandising. Bad photos and missing variants still kill conversion. Recommendations amplify what the catalog already is.
Confusing recommendations with the page agent. One suggests products. The other can navigate filters and cart steps for the visitor.
Turning on every Generatify AI surface at once. Isolate the shop placements until they look trustworthy.
Practical checklist before you call it done
- Connect a Generatify Pro model key if you have not already.
- Enable product recommendations.
- Switch on product, cart, and thank-you placements you actually want.
- Verify thank-you templates still run WooCommerce hooks.
- Clean obviously bad best sellers (test SKUs, discontinued items).
- Run cold-visitor, browser, and buyer-path scripts.
- Confirm recommendations change with browse history rather than staying frozen.
- Keep checkout unblocked on cart (suggestions should not trap the button).
- Leave helpdesk, page agent, and editorial AI for a separate session.
- Revisit after a week of real traffic and prune any placement that underperforms visually in your theme.
Keep the catalog honest too: clear titles, real stock status, and decent images give the recommendation engine something worth suggesting. Behavioural ranking cannot rescue a neglected product record.
Takeaway
WooCommerce does not need another static upsell row. It needs suggestions that follow what people browsed and bought, with a honest fallback when a visitor is new.
Generatify builds those recommendations from real browse and order history, places them on product, cart, and thank-you pages, and falls back to best sellers on cold start. Connect your own model key. Enable the three placements on purpose. Test with a private window and a returning path before you call merchandising “done.”
One Pro membership unlocks Generatify plus Progressify, Lightify, Rankify, and Snapshotify. New Pro plugins join existing memberships. Plans differ by site count only (Pro $37.99 for 1 site, Pro Plus $74.99 for 5, Pro Max $138.99 for 100), with a 14-day money-back guarantee. See pricing when related products should come from behaviour instead of a spreadsheet.
Turn the placements on. Click through a real cart once. When the next product feels earned rather than pasted, the recommendation row finally belongs on the store.