Troubleshoot suggestions
When to use this: the Counsellor isn’t behaving as expected. Most issues are data or timing, not a broken model.
”I don’t see any suggestions.”
Most likely the add-on isn’t enabled for your pharmacy, or products aren’t classified yet. Confirm the add-on is on your plan (Availability & permissions) and that the product is in scope (in stock, or sold in the last 90 days). A freshly added / restocked product is classified within the hour.
”There’s no trade-up ribbon on this item.”
Either the item isn’t classified yet, or no in-stock alternative has a higher margin — so there’s genuinely nothing to trade up to. The ribbon only appears on a real margin gain.
”It suggested a more expensive product.”
Up-sell ranks by margin, not price — a higher-priced line can earn more margin (the ribbon shows “+Rs N margin for you”). The customer is only charged more if you tap Upgrade.
”A substitute looks like the wrong medicine.”
Flag it. Misclassified products are usually vetoed automatically when their recorded ingredients disagree with the AI; flagging gets it reviewed and suppressed. The pharmacist at the counter is always the final clinical check.
”It offered a suppository / injection as a swap for a tablet” / “mixed up SR with IR.”
It shouldn’t — route/form and release are locked. Report any case you find; it indicates a classification issue on that product (this matters clinically, e.g. carbamazepine, nitroglycerin). See How it stays safe.
”I reported a suggestion but my colleague still sees it.”
By design. To guard against a single mistaken flag, a suggestion is removed pharmacy-wide only after two different cashiers flag the same pair — or an admin upholds it in the Pharmacist Center. It disappears from your screen immediately.
”Cross-sell only shows syringes / saline / gloves.”
Old behaviour, since corrected (ubiquity penalty + thin-sample correction). If staples still dominate, the cross-sell cache may not have rebuilt yet — it refreshes nightly per tenant.
”Suggestions seem inaccurate across the board.”
Usually data quality, not the model. Fill in cost price (drives up-sell margin ranking) and dosage form / strength where possible — the engine degrades gracefully on missing fields but gets noticeably more accurate as they’re filled in. See Data requirements & thresholds.
”I just enabled the module / imported a big catalogue — nothing yet.”
Classification runs in the background at ~1,500 products/hour — a 3,000-SKU pharmacy is covered in ~2 hours. Products already seen elsewhere on the platform classify near-instantly via the shared cache.