Your POS tells you what sold yesterday. It doesn't tell you why store two beat store four, whether the webshop is feeding your shops or eating them, or which supplier's margin survived the sales. At some point someone says "we need BI." This guide is for that moment.
It covers whether you need it at all, what a working setup is actually made of, the five routes on the market, what each one costs once you count everything, and the questions to ask before you sign. It is written to be forwarded: if you have to convince a partner, an owner or a finance lead, the tables and the checklists are the parts to send.
Do you need BI at all?
If you have one store, one POS system and one sales channel, probably not yet. Your POS reports plus a spreadsheet cover it, and you know which lines are moving because you walk past them. A BI tool here adds a subscription and a login without adding a decision.
Site count is the easy test, but it isn't the only one. Four things push a retailer into needing more than the till can show, and any one of them can apply to a single large shop:
More than one site. The question changes from "how did we do" to "why did that store do better than this one." That comparison is what POS reporting isn't built for.
More than one data source. Shops and a webshop on different systems, or sales that only make sense next to staff hours, footfall counters or your accounting.
A complicated range. Thousands of SKUs, sizes and colours, several brands or categories with different margins.
An unusual trading pattern. Most of your year happening in a few weeks, a fiscal year that doesn't start in January, or sales in more than one currency.
Before you look at any product, answer two questions in writing. What do you want to see that you can't see now? "Better insight" isn't an answer; "sell-through by size across all stores every Monday" is. And who will open it every week? Name the person. A reporting tool bought without someone who owns it ends up as a login nobody uses, whichever route you pick.
What a BI setup is actually made of
Every working setup has two layers, and most buyers only price the first.
The reporting layer is what you see: the dashboards, charts and the Monday report. This is what Power BI, Tableau and Qlik sell, and it's what their price pages show.
The data layer sits underneath. It pulls sales out of your POS, webshop and other systems every day, cleans them up, matches products and stores across systems, and stores years of history so the reports can be fast. It also needs your definitions built in: what counts as margin, which stores are like-for-like, when your financial year starts. Without this layer the reporting tool has nothing to show.
The data layer is where most of the work and most of the cost lives, and it never finishes. Your POS updates and renames a field, you add a supplier, you open a store, and something downstream has to be adjusted or a report quietly breaks. When a nightly data load fails, someone has to notice and fix it.
So when you compare options, the real question is: who builds the data layer, who looks after it, and is that included in the price you're being shown?
The five routes
There are five ways retailers get to cross-store reporting. The difference between them is mostly who builds and maintains the data layer.
1. Spreadsheets and a person. Weekly exports from the POS, built into a workbook by whoever is good at Excel. Cheap and immediate. It works until that person is on holiday, gets busy, or leaves.
2. Build it with a general BI tool. Buy Power BI, Tableau or Qlik licences, then have someone build the data layer and the reports: your own staff if you have the skills, otherwise a consultancy. These tools are genuinely the most capable and flexible option. If someone in your team already knows one and has the time, building is a real choice. The catch is that the capability only exists once someone has built it, customising usually means learning the tool's own formula and query languages, and every change after launch goes back to whoever built it.
3. Buy a retail-specific analytics service. A ready-made product that connects to common POS systems and arrives with retail reports already built: sell-through, margin, stock, like-for-like, store comparison. The data layer is the vendor's job. Faster to get going and easier for store managers to use without training. The trade-off is flexibility: you can't build anything the vendor didn't anticipate, and connecting non-retail sources may not be possible.
4. Hybrid. Two versions. Either a ready-made retail service with the vendor adding custom integrations or reports on top, or head office runs a general BI tool for finance-style analysis while the stores use a ready-made service. In practice the line between build and buy is blurring: general tools ship more prebuilt connectors every year, and ready-made services increasingly let users create their own metrics without code.
5. AI. The new question everyone asks: can't I just ask an AI to build this? As of late 2026, not on its own. AI can help you analyse data that is already clean and connected, and AI assistants are being added to every tool on this list. What it can't yet do reliably is build and run the data layer: the daily integrations, the history, the definitions. Treat AI as a feature to look for inside the other routes, not a route by itself. Check again in a year.
Spreadsheets | Build with general BI | Buy retail-specific | Hybrid | |
|---|---|---|---|---|
Who builds the data layer | Your Excel person, weekly | Your team or a consultancy | The vendor | The vendor, plus your team or consultancy for the extras |
Who maintains it | The same person | Whoever built it, usually billed | The vendor, in the subscription | Split; agree in writing who owns what |
Right for you if | Under five sites, stable questions, someone with time | You have or will hire a data person, or your questions are unusual | You want standard retail questions answered soon, and nobody on staff will build reports | You need retail basics fast plus one or two unusual sources or reports |
Store managers can use it | If they're sent the file | After training, usually view-only | Usually yes, with a walkthrough | Depends which part they use |
Main cost | Hours every week | The build and upkeep; licences are the smaller part | The monthly fee | Both |
What you give up | Consistency and the ability to ask a new question quickly | Time before it's useful, and dependence on the builder | Flexibility beyond retail | Simplicity: two suppliers or two tools |
Sign you chose wrong | The workbook is late two weeks running | Months in, it's still being built | You keep asking for something and hearing "it's on the roadmap" | Nobody knows which number is the right one |
What it really costs
The licence is the number on the website. The data layer is the number that isn't. For a built solution, building and maintaining the data layer usually costs more over the first few years than the licences do, and no vendor publishes it because it depends on your systems.
Published prices, and what's inside them
List prices, checked September 2026. Verify on each vendor's own page before you sign; they change.
Option | List price | Includes | Doesn't include |
|---|---|---|---|
Power BI Pro | $14 per user per month | Building and viewing reports, sharing inside your company, data models up to 1 GB, up to 8 scheduled refreshes a day | The data layer, retail reports, connection to your POS beyond generic connectors, building, upkeep |
Power BI Premium Per User | $24 per user per month | As Pro, plus larger models, up to 48 refreshes a day and more AI features | As Pro. The extra buys capacity, not content |
Power BI capacity (Fabric) | From about $262 a month; the tier where viewers need no licence costs roughly $5,000 to $8,400 a month | Dedicated computing power; at the top tier, unlimited free viewers | Below that top tier, every user still needs a Pro licence |
Tableau Cloud Standard | $75 per Creator, $42 per Explorer, $15 per Viewer, per user per month, annual contract | Building, publishing and viewing dashboards in the cloud. At least one Creator required | The data layer, retail reports, building, upkeep. Some security features cost about 50% more per seat on Enterprise |
Qlik Cloud Starter | $300 a month for 10 users and 10 GB of data, annual | Dashboards, standard connectors, report scheduling | More data capacity. Above 10 users you move up a tier |
Qlik Cloud Standard | $825 a month for 25 GB of data, unlimited users, annual | Unlimited users, which is rare in this category | The data layer, retail reports, building, upkeep. Price rises with data volume, so years of SKU-level history cost more |
Zoined (retail-specific) | €99 per site per month at 1 site, €68 at 2 to 5, €48.50 at 21 to 35, €34 at 71 to 100. One-off setup €200 to €900. 10% off if paid annually; otherwise monthly with no minimum term | Integration with 38 certified POS and ERP systems, the data layer, prebuilt retail dashboards and reports, budgets and forecasts, custom metrics without code, onboarding, training, support and unlimited users. 30-day free trial on your own data | Reports outside retail and restaurant. Setup is quoted separately if your POS isn't on the certified list. The base rate covers up to 2 POS terminals per site; more terminals per site cost more. SSO is extra |
Three retailers, priced on licences alone
Assumed users: the owner, one buyer or operations lead, and one manager per store. Change the numbers to fit you. Prices are shown in the currency each vendor lists in.
4 stores, 6 users | 11 stores, 13 users | 25 stores, 27 users | |
|---|---|---|---|
Power BI Pro | $84 a month | $182 a month | $378 a month |
Tableau Cloud Standard | $150 a month (1 Creator, 5 Viewers) | $255 a month (1 Creator, 12 Viewers) | $492 a month (1 Creator, 1 Explorer, 25 Viewers) |
Qlik Cloud | $300 a month (Starter) | $825 a month (Standard) | $825 a month (Standard) |
Zoined | €272 a month (4 × €68), plus setup | [11 × the 6 to 20 site rate, confirm from calculator], plus setup | €1,212.50 a month (25 × €48.50), plus setup |
On this table alone, building looks cheaper. That's the point of the next part: for the three general tools, every cell above is missing the data layer, the build, the upkeep and the storage. For Zoined, those are already inside the fee. Compare like with like.
The costs that aren't on the price page
When you build with a general BI tool, budget for each of these separately:
Data storage. The warehouse your history lives in. Often forgotten until after the licences are bought.
The build. Connecting your POS and webshop, modelling the data, defining the metrics, building the reports. Billed by the day if a consultancy does it.
Upkeep. Fixing what breaks when a source system changes, adding a new store or supplier, building the report someone asks for next month. General BI vendors don't fix your data model for you.
Monitoring. Someone noticing when last night's data didn't load, and fixing it before the Monday meeting.
Training. Building reports in a general tool usually means learning its query and formula languages.
Growth. With per-user pricing, every store manager you add is another licence. Teams then quietly decide some managers don't need to see the numbers.
One more thing for whoever signs off the budget. A build is often booked as an upfront investment (capex), while a subscription is an operating cost (opex). Depending on how your business is measured, that changes how each option looks on paper, even when the total over three years is the same. Ask your finance lead which matters for you before comparing quotes. And check whether a build quote includes the upkeep after launch: it often doesn't.
Three years, everything counted (estimate)
This puts the hidden costs back in. Nobody can quote your build without seeing your systems, so the figures below rest on stated assumptions. Swap in your own; the method matters more than our numbers.
Assumptions: a consultancy at €1,000 a day builds the data layer and core reports in 30 days, then keeps it running for one day a month. Data storage at €100 a month. Spreadsheet time valued at €35 an hour, at 5, 10 and 20 hours a week as the business grows. Dollar prices converted at $1 = €0.90. Zoined at list price with the maximum €900 setup fee and no annual discount.
3-year total, € | 4 stores, 6 users | 11 stores, 13 users | 25 stores, 27 users |
|---|---|---|---|
Spreadsheets and a person | 27,300 | 54,600 | 109,200 |
Power BI Pro, built by a consultancy | 72,300 | 75,500 | 81,800 |
Tableau Cloud Standard, built by a consultancy | 74,500 | 77,900 | 85,500 |
Qlik Cloud, built by a consultancy | 79,300 | 96,300 | 96,300 |
Zoined | 10,700 | between 20,100 and 27,800 | 44,600 |
For each built option, €69,600 of the total is the build, upkeep and storage; the rest is licences. That is why the licence-only table above is misleading on its own.
Three things change the picture. If you already employ someone who can build and maintain the data layer, the build and upkeep become their time rather than an invoice, and building gets much cheaper. If your build is leaner than assumed, the gap narrows fastest at larger sizes: halve the build and upkeep and Power BI at 25 stores comes to about €47,000, close to Zoined's €44,600. And spreadsheets look cheap only until you price the hours; at 11 stores they cost more than a packaged service.
Rough estimate for comparison only. Your figures will differ; ask for fixed-price quotes using the checklist below.
Questions to ask before you sign
Ask any vendor
How long from signing to the first report we'd actually use, and how did you measure that? A range with no method behind it is a guess.
What does it cost when we open a store, or add ten users? Per-user and per-site pricing behave very differently as you grow.
Who builds a new report, and what does that cost? If the answer is "you do," make sure someone on your team can.
What's included in support? Specifically: who fixes it when a data load fails or our POS system changes?
Does it handle our calendar and our currencies? A financial year that starts in April, weeks that run Monday to Sunday, sales in more than one currency.
What happens to our data if we leave? Get the answer in writing.
Show us a retailer our size who has been live for a year. A name and a store count, not a logo wall.
If you're building, ask two consultancies for a fixed-price quote
Refuse a day-rate estimate. Ask each to price these six things separately:
Connecting your POS, webshop and any other sources, including the data model
Data storage, with the monthly cost
A named list of reports, at minimum the questions you wrote down at the start
Monitoring and fixing failed data loads
A monthly retainer for changes: new stores, suppliers, categories, report requests
Training for the people who will actually use it
The gap between the two quotes will tell you more than any figure in a buying guide.
The short version
One store, one POS, one sales channel: your POS reports and a spreadsheet are probably enough.
Several shops, or several systems, and nobody to build reports: a retail-specific service gets you cross-store numbers fastest, at a price you can see upfront.
A data person on staff, or questions nobody else has: building with a general BI tool gives you the most control. Budget for the data layer and the upkeep, not just the licences.
In between: a hybrid, with one clear owner for each part.
The mistake in every direction is the same: choosing the tool before naming the question and the person who will look at it.
Zoined is an analytics service built for retailers and restaurant operators. It connects to 38 POS and ERP systems, arrives with retail reports already built, and is priced per site with unlimited users. See Zoined's pricing or start a 30-day free trial on your own data. If you buy through your POS provider, your price comes from them.
Editor notes, remove before publishing
Prices for Microsoft, Tableau and Qlik are from secondary sources checked September 2026. Verify each on the vendor's own pricing page, add the links as sources, and date them.
No implementation durations or failure rates are in the guide. The episode says 6 to 12 months; internal docs say 3 to 7 weeks and 2 to 4 months; none is sourced. The guide argues from who builds and maintains instead.
"Licences are the smaller part" is stated as the mechanism, not a figure. If Sasa can source a ratio, it becomes a much stronger line.
Zoined's 11-store cell is a placeholder: the pricing FAQ has no 6 to 20 site band. Get the rate from the calculator. Also confirm the footer pricing URL, and that the partner is happy with list prices shown.
Customer quotes were left out of this version to keep it neutral and short. Candidates if cleared: Work World (per-user licences), Didriksons (40 minutes to 10), Lab9 (ran Power BI before, if permission comes through).
Episode 3 tie-in: the episode names Zoined several times. Check with the partner before linking it from a co-authored guide.
Zoined is named in the two price tables and the footer. The five routes and the summary stay generic.
TCO table assumptions (€1,000 day rate, 30-day build, 1 day a month upkeep, €100 a month storage, €35 an hour spreadsheet time, $1 = €0.90) are ours and unsourced. Sasa should sign them off before publishing; they are deliberately stated so readers can replace them. Zoined's 11-store figure is a range between the 2 to 5 and 21 to 35 site rates until the 6 to 20 band is confirmed.
