Delete the report before you automate it: the test most owners skip

Roughly half the recurring work that owners ask me to automate is output nobody uses. Not badly-made output. Not output that needs a better format or a nicer dashboard. Output that no human being reads, acts on, or would miss if it vanished.

The daily sales summary that goes to a WhatsApp group where the last reply was months ago. The weekly stock report emailed to a partner who checks the actual stock by walking into the godown. The monthly MIS deck assembled by an accountant over two full days, opened once, skimmed, and filed.

When an owner says "automate this report", the honest first question is not "which tool" or "what will it cost". It is: does anyone use this? And there is a much cheaper way to answer that question than paying someone like me.

Stop sending it. See who asks.

Why dead reports pile up

No one sets out to create a useless report. Every dead report was born alive.

There was a crisis once. A distributor's payment slipped, or a stockout embarrassed you in front of a customer, or a partner suspected leakage. So you asked for a report. For a few weeks you read it closely, because the wound was fresh. Then the crisis passed, the reading stopped, and the sending continued. The person compiling it has no authority to stop, and stopping feels like admitting the work was pointless. The person receiving it will not say "stop sending this", because that sounds like they are not on top of things.

So the report survives on politeness. In a business run on WhatsApp and Excel, it can survive that way for years, quietly eating a few hours of someone's week.

Automation consultants love these reports, and you should be suspicious of that. A recurring, well-defined, repetitive task is the easiest thing in the world to sell an automation for. The pitch writes itself: your staff spends hours every week on this, the system does it in seconds. Everything in that pitch can be true, and the project can still be worthless, because the hours saved were being spent producing something nobody wanted. You have not automated work. You have automated waste, and you have paid to make the waste permanent. A dead report that a person compiles will eventually die when that person leaves or pushes back. A dead report generated by a script will run forever, because software never gets bored and never asks why.

The two-week silence test

Here is the whole method. It costs nothing and needs no tools.

Pick one recurring report. Tell the person who compiles it, privately, to pause it. Do not announce anything to the recipients. No "we are reviewing our reporting", no "please confirm if you still need this". Announcements poison the test, because people asked "do you still want this report?" almost always say yes. Saying yes is free and saying no feels careless. You are not testing what people say they want. You are testing what they notice is gone.

Then wait two weeks and watch for three outcomes.

Someone asks within days. Good. The report is load-bearing. Find out what they actually do with it, because the answer tells you what the report should contain, which is usually less than it currently contains. This report is a genuine automation candidate.

Someone asks after a fortnight, vaguely. "Were we not getting that sales thing earlier?" This report is a habit, not a tool. The asker noticed the absence, not the missing information. Ask them what decision they take from it. If the answer is a shrug, let it stay dead.

Silence. The most common outcome, in my experience, and the most valuable. You have just found out that a slice of someone's week was being spent on nothing. Delete the report. Do not automate it, do not improve it, do not move it to a dashboard. Reassign the hours.

The test works because it measures behaviour instead of opinion. Everyone believes they read their reports, the same way everyone believes they will start exercising in the new year.

A worked example: the distributor outstanding report

A trading business selling through distributors typically has someone in accounts preparing a party-wise outstanding statement. Every week, the accountant pulls balances from Tally, formats them in Excel, and mails the sheet to the owner and the two people who handle collections.

An owner in this position usually asks me for the obvious build: pull from Tally automatically, generate the sheet, push it to WhatsApp every Monday morning. It is a reasonable request and a straightforward build.

But run the silence test first and you often find something more interesting. The collections people do not open the sheet. They already know their top overdue parties by name, because they are on the phone with them every day. The owner opens it, but only scans one thing: which parties crossed the credit period this week. Fifty rows of data, one question being answered.

The right system, then, is not an automated version of the weekly sheet. It is an exception alert: a message only when a party crosses its credit limit or its balance jumps past the agreed period. No news means no message. That is a smaller build than the one requested, it gets read precisely because it arrives rarely, and it only became visible because the original report was allowed to die first. Automating the fifty-row sheet would have worked flawlessly and helped nobody.

The same pattern shows up elsewhere. During GST filing season, staff assemble reconciliation summaries that the CA's own software already produces; the internal version exists because it always has. Daily "enquiries received" counts from the WhatsApp business number get compiled by hand into a message the owner stopped reading once volumes stabilised. The test is the same in every case. Kill it quietly, watch the silence, and only build for what someone actually came looking for.

What survives the test still is not automatic

A report that people ask about has earned a closer look, not an automatic build. Three follow-up questions, in order.

What decision does it drive? If the honest answer is "awareness", be careful. Awareness is what people say when a report changes nothing. A report worth automating has a verb attached: call this party, reorder this item, question this expense.

Is it needed at this frequency? Reports drift towards daily because daily feels rigorous. Most decisions in a mid-sized business are weekly or monthly decisions. A daily report feeding a monthly decision is noise with a schedule.

Is it a report at all, or an alert wearing a report's clothes? This is the distributor example again, and it is the most common finding. People do not want data on schedule. They want to be told when something needs them. If the reader scans forty rows to check that nothing is wrong, what they want is a system that stays silent until something is wrong.

Sometimes the answers genuinely justify the original report, and then automating it is a fine project. But you reach that conclusion having tested it, not having assumed it.

And say the quiet part plainly: when a report fails the silence test, the correct amount to spend automating it is nothing. There is no cheap version of a system nobody needs. I have declined builds on exactly these grounds, and it is a strange conversation, because the owner arrived ready to spend and leaves having deleted a task instead. But an automation that produces unread output is worse than the manual version, because the manual version at least had a human in it who might one day ask why.

The uncomfortable version of this rule

The report is the easy case, because a report is visibly output. The harder version applies to any recurring task: approval steps where the approver has not rejected anything in living memory, registers maintained in parallel with software that already records the same thing, photographs of documents sent daily to a folder nobody opens.

Before automating any recurring work, ask what happens if it simply stops. Frequently the answer is nothing, and "nothing" is the finding. The best outcome of an automation review is often a shorter list of things the business does at all. Fewer tasks, then automate the survivors. In that order, always, because automating first freezes the waste in place.

This week

Pick the one recurring report you are most confident people rely on. Not the one you suspect is dead, the one you would defend. Tell the person who prepares it to pause it, tell no one else, and put a note in your calendar for two weeks out.

If your phone rings about it, you have found real work, and you now know exactly who uses the report and for what. Automate that, and only that.

If your phone stays silent, you have just discovered what half my client conversations discover: the cheapest automation is deletion, and you did not need a consultant for it.