You can answer "what is happening on my shelves?" four ways. We've laid them out side-by-side, with no marketing spin. The fourth column is honest about where we still have work to do.
| Spreadsheets | Legacy BI (Tableau, Power BI, Qlik) | Build it yourself | + Hank | |
|---|---|---|---|---|
| Time-to-answer | ||||
| Data freshness | 3–14 days. Whenever the analyst pulled the file. | 1–24 hours. Bound by the ETL schedule. | Whatever you choose to build. Realistic for in-house: 1–6 hours. | < 60 seconds. Streaming POS into the warehouse, into Hank. |
| Asking a new question | 30–120 minutes. New tab, new pivot. | 2–48 hours. Open a ticket with the BI team. | 2–48 hours. Same - your BI team, your queue. | 3 seconds. Plain English to Hank. |
| Plain-English answers | No. You write the formula. | No. You write the SQL or the DAX. | Only if you've built an LLM layer (most haven't, and it's hard). | Yes. Every answer cites its source rows and gives a confidence score. |
| Cost-to-stand-up | ||||
| Initial setup time | Hours. Already on every laptop. | 3–9 months. Schema work, ETL, modeling, dashboards. | 9–18 months. Plus you keep maintaining it. | 2–4 weeks. POS integration → live. |
| Annual cost · 50-store retailer | $0 in software. ~$280K in unaccounted analyst time. | $60K–$180K licences + $400K–$800K in BI staff and contractors. | $1.2M–$2.4M loaded cost. Three engineers, one PM, infrastructure. | $210K all-in. Subscription + onboarding. No BI team required. |
| Maintenance burden | One analyst leaves and the model breaks. Always. | Continuous. Schema changes, new SKUs, dashboard rot, version upgrades. | Continuous. It is now a product. You own a product. | Ours, not yours. POS schema changes are absorbed by tapestry's integration team. |
| What it can do | ||||
| Cross-store comparisons | Possible. Painful at > 10 stores. | Yes - if the model was built for it. | Yes - if you built it that way. | Yes. Ranked, ranked-by-impact, ranked-with-actions. |
| Recommendations & actions | No. Data only. | No. Data only. | Only if you built a recommendation engine. Few have. | Yes. Hank surfaces the next action - restock, reset, reorder - ranked by margin impact. |
| Supplier collaboration | Emailed spreadsheets. | Rarely shared outside the company. | Custom build - supplier portals are their own product. | Built in. Supply+ gives suppliers a governed lens on the same data, with your consent. |
| POS integrations | Manual. Whichever export the team can write. | Often a connector exists; quality varies. | You build them. All of them. Forever. | 7+ certified POS partners. Read-only, no writes, never. |
| Trust & governance | ||||
| Data residency | On the analyst's laptop. | Wherever you host it. | Wherever you host it. | Your region only. AU, US, EU instances. SOC 2 Type II in flight Q3 2026. |
| What suppliers can see | Whatever you email them. | Not designed for cross-org sharing. | Whatever you build. Usually too much, or too little. | You control, SKU by SKU. Suppliers see only what you toggle on. Logged, reversible. |
A 30-minute walkthrough with your own SKUs. We'll show you Hank against a sample of your shelf data and you decide if the comparison above holds up.
No software is universally better. Here is where each of the alternatives is actually the right call.