Top 10 buyer questions on AI automation

The difference is between a tool and a system. A chatbot helps one person work faster. An automation changes how the business runs: stock alerts that trigger themselves, franchisee reports that compile overnight, review replies drafted for every branch.

Do it yourself if:

• You're automating one or two personal tasks
• Your systems don't need to talk to each other
• Someone on your team has time to learn and maintain it

Bring in help if:

• The workflow crosses two or more systems (for example POS to accounting)
• It touches customer or staff personal data, which brings in POPIA or UK GDPR
• A failure would cost you money, like a missed order or a wrong stock count
• You've already tried and the tools sit unused

A good middle path: pay for a short scoping engagement, then decide what to build in-house and what to outsource.

What to watch: Zapier gets expensive at high volume because every step counts. Make's credits run out fast on scenarios with many steps. n8n is free to self-host, but someone has to run, secure and back up the server.

All three bill in US dollars or euros, so your cost moves with the exchange rate. That matters for South African businesses budgeting in rand.

The tool matters less than the design. A badly planned workflow is costly and fragile on any platform.

Three things drive the price: how many systems need connecting, how much personal data is involved, and whether the consultant trains your team or just builds and leaves.

Ask for a fixed price per phase rather than open-ended day rates. Then budget for the software subscriptions on top of the consulting fee.

Ranges are drawn from published 2026 market guides and agency listings. Figures vary widely; treat them as a guide, not a quote.

ROI in restaurants and franchises usually comes from four places:

• Admin hours saved: reports, reconciliations, rosters and data entry that managers do by hand
• Fewer errors: wrong stock counts, missed orders, mismatched invoices
• Less waste: better stock forecasting means less spoiled food
• More repeat revenue: faster review replies, loyalty follow-ups, win-back messages

A simple way to work it out (illustrative): if 10 branch managers each spend 3 hours a week compiling reports, that's about 130 hours a month. Automate it, and multiply the hours saved by their hourly cost. Compare that to the build fee plus monthly software cost.

Measure the baseline before you start: hours spent, error rates and waste. Without it, nobody can prove the ROI afterwards.

A good first-wave shortlist:

1. Franchisee and branch reporting: daily sales and KPI packs compiled automatically from the POS
2. Stock and ordering alerts: low-stock warnings and reorder suggestions before a store runs out
3. Review management: Google and delivery-app reviews collected in one place, with draft replies
4. Invoice and supplier admin: supplier invoices captured and matched to orders
5. Customer follow-up: loyalty, birthday and win-back messages triggered from order data

Avoid starting with anything customer-facing and high-risk, such as a fully automated complaints bot. Prove value on an internal process first, then expand.

The test for every candidate: how many hours a month does it take, how often does it go wrong, and what does an error cost?

Work through it in this order:

1. Map your systems: list every tool (POS, delivery platforms, accounting, payroll, loyalty) and what data each holds
2. Use native connectors first: many POS and accounting systems already sync sales and payments
3. Check for APIs: systems without native links often have an API an automation platform can use
4. Automate the gaps: use Zapier, Make or n8n for the flows that aren't covered
5. Add one reporting view: pull everything into a single dashboard for head office

The common failure is building point-to-point links with no owner and no error alerts. Every automated flow should have a named owner and a notification when it breaks.

South Africa (POPIA):

• Most AI and automation tools store or process data overseas. Section 72 of POPIA only allows that if one of its grounds applies, such as an adequate foreign law, a binding agreement, the person's consent, or it being needed to perform a contract.
• In practice, that usually means a data processing or transfer agreement with each vendor.
• Section 71 limits decisions made solely by automated means that significantly affect a person.
• Fines under POPIA can reach R10 million.

UK (UK GDPR and the Data (Use and Access) Act 2025):

• The 2025 Act relaxed the rules on solely automated decisions, but safeguards remain: transparency, a way to challenge the decision, and meaningful human review.
• Stricter limits still apply to special category data, such as health information.

For any business: don't paste customer or staff personal data into public AI chat tools, run a privacy impact assessment before go-live, and keep a record of which vendors hold what data and where.

This is general information, not legal advice. Check specific cases with a data protection lawyer.

1. Which restaurant, retail or franchise businesses have you automated, and what changed in measurable terms?
2. What will you automate first, and why that process?
3. How will we measure success, and what's the baseline?
4. Is the price fixed per phase, or open-ended?
5. Which tools will you use, and who owns the accounts and logins?
6. Where will our customer and staff data be stored, and how do you handle POPIA or UK GDPR?
7. What happens when an automation breaks? Who gets alerted, and how fast is it fixed?
8. Will you train my team, or will we depend on you?
9. What's included after launch, and what does ongoing support cost?
10. Can I speak to a past client?

Watch for:

• "We'll automate everything": good partners start with one process and prove it
• Guaranteed ROI or headcount cuts before they've seen your numbers
• Tool-first selling: every problem gets the same platform or chatbot
• No questions about your data: no mention of POPIA, GDPR or where data goes
• Accounts in their name: you should own every login, workflow and subscription
• No documentation or handover plan
• Open-ended day rates with no cap or phase gates
• No case studies or references from businesses like yours
• Very low rates: a price far below the market often means a template that won't fit your systems

Score yourself on five areas:

If you're not ready in two or more areas, fix those first. Automating a messy process just makes the mess faster.