Which Workflow Should You Automate First?

Most teams automate whatever is shouting: the inbox that exploded this week, the report someone asked for on Friday, the one customer who complained loudly. That feels responsive. It rarely picks the work that would have paid back fastest.

The better question is which piece of work is frequent, slow, predictable, and expensive all at once. The free Automation Opportunity Quiz is five questions built to force that comparison. It will not uncover a hidden process. It will stop you starting with the wrong one.

What the quiz is for

It is a free discovery tool. You answer five questions about one workflow, or about whichever mess is worst, and you get a score, a plain label telling you whether this is a strong first candidate, a recommended workflow, a short note on what an agent would actually do, and a link to the matching solution.

It earns its keep when leadership agrees you should use AI and nobody agrees where. It is not an audit, and it is not a substitute for pricing the work once you have a candidate.

What it weighs

The first question names your biggest operational challenge. That answer decides the recommendation: support overload points to customer support, slow follow-up points to lead qualification, missed appointments point to appointment booking, and manual data or reporting work points to internal operations.

The remaining four questions decide the score. Volume asks how often this happens. Handling time asks how long one instance takes a person from start to finish. Predictability asks whether the same inputs produce the same outputs. Impact asks what this workflow is worth in revenue or cost each month.

Predictability carries real weight, and that is deliberate. High volume without stable rules is still a people problem, and adding an agent to it usually moves the mess rather than removing it. Most of the value from AI in day-to-day operations shows up on exactly this kind of repeatable, high-frequency work (McKinsey's research on where generative AI value lands).

How to answer honestly

Pick the challenge that is eating the week, not the one that presents well. If support is fine and enquiries are rotting, pick enquiries.

Count only this workflow when you answer volume. If you chose support, that is tickets, calls, and chats, not orders.

For handling time, use the average for one instance, including the wrap-up if a person always does wrap-up.

On predictability, let the most sceptical person in the room answer. If they say the work is rarely predictable, believe them. That is not a failure, it is an instruction to tidy the process first.

For impact, use the revenue or cost tied to this workflow alone. The bands are shown in both rupees and dollars so you do not have to convert anything in your head.

Then read the breakdown, not only the total. If the score is high because of volume and low on rules, you have a different project than you thought.

Using the result

Follow the recommended solution link to see how that agent family works, then move to the AI Agent ROI Calculator and price the same workflow properly. You will enter volume and time again, which is healthy, because entering them twice is where people notice their first guess was optimistic.

If support wins, the voice cost estimator usually tells you more. If enquiries win, run the speed-to-lead calculator. If bookings win, run the no-show calculator.

You can retake the quiz for a second workflow. Write both scores down and compare. Retaking the same workflow until the label improves is just an expensive way to lie to yourself.

Who this suits

It suits you if you have been asked to pick a first agent without a map: an owner with three fires, an operations or support lead, a sales ops manager, or a consultant scoping a client. Run it with the client in the room if that is your situation. The useful argument almost always happens on the predictability question rather than on the final number.

Skip it if the first workflow is already obvious and you only need a rupee figure. Go straight to ROI.

Two mistakes are worth naming. A strong score is not a promise about how many tickets will be handled without a person. And a low score is the tool doing its job, not a reason to press ahead anyway.

Take the free Automation Opportunity Quiz, then price that workflow, start free, or book a strategy call.

Questions worth settling before you build

A score tells you where to look. These four questions decide whether the build goes well.

Who owns the outcome? Not who signs the invoice. Who checks, on a Friday, whether the work actually completed. Automation without an owner drifts within a month.

What does the agent need access to? Almost every useful workflow touches a calendar, a CRM, an order record, or a ticket queue. If nobody can grant that access in the next two weeks, pick a different first workflow, however good the score.

Where does a person take over? Write the rule down in one sentence before anything goes live. "Anything about a refund over a set amount goes to a person" is a rule. "It should use judgement" is not.

What will you measure in week one? Pick one number: replies inside five minutes, calls answered after hours, bookings confirmed without a staff call. One number you already track beats a dashboard nobody opens.

Reading a low score properly

A low score is not a verdict on your business. It usually means one of three things.

The volume is real but the rules are not, so the same request arrives in five shapes because nobody has agreed on the process. Fix the process on paper first, then automate what is left.

Or the work is predictable but rare, in which case automating it is a hobby rather than a project. Note it and move on.

Or the value is unclear because nobody has priced the workflow yet, which is what the ROI calculator is for. Score again once you have the figure.

What the first workflow usually is

In practice, first workflows cluster. Businesses with a phone line start with answering and routing calls after hours. Businesses with a form start with the first reply and qualification. Clinics and studios start with confirmations and rescheduling. Back-office teams start with the status question they answer forty times a week.

None of those are ambitious. That is why they work. The first agent's job is to prove the boundaries hold and the handover is clean, so the second one is a conversation about scope rather than about trust.

If you want the wider context on why we sequence work this way rather than selling a platform rollout, why we built 20X covers it. If the quiz says you have several leaks rather than one, the Human Work Tax diagnostic will rank them for you.

What happens after you pick

Once the workflow is chosen, the sensible order is small and boring. Write the rule for what the agent may do. Give it the knowledge you have already approved, rather than everything you own. Test it against last month's real conversations, including the awkward ones. Launch on one channel. Read the transcripts in week one, and again in week four.

Whichever package you use to get there, the cost splits the same way: the fee pays for the people who design and build it, and tokens pay for the platform running it afterwards. Explore at ₹4,999 is a single discovery session, Start at ₹25,000 plus ₹9,999 a month is where a team gets involved and your first workflow goes live, and Transform and Scale exist for wider programmes. You can also do all of it yourself on usage alone.