How to measure software adoption without relying on logins

A login proves a password works. It says nothing about whether the person raised a quote, posted a receipt, closed a job card or entered a single stock movement all week. Someone can sign in at nine, glance at the dashboard, then spend the rest of the day inside a spreadsheet, and the login chart counts that person as active.

Vendors like the login chart. It climbs easily. Owners should distrust it for that very reason. You pay for work done inside the software, and the record is where that work leaves its mark, so read the records rather than the sign-in log.

So stop asking who logged in. Ask what the data looks like on a Thursday afternoon, when the week’s work should be half done and half recorded in the system. Open the quotation register, the GRN list and the open job cards, then read them the way an accountant reads a ledger, line by line.

In a CRM, count open deals that carry both a next step and a dated follow up, then divide by all open deals to see how many are alive. A pipeline full of those means salespeople keep their real work in a notebook, a private spreadsheet or a WhatsApp thread that the manager never sees.

Check when each deal was last touched. If most deals were last touched three weeks ago while reps claim to be busy, the calls and emails are happening off the system.

Pick five deals at random. Ask the rep to explain each one. Allow the screen only. A rep who reaches for a phone or a diary shows you where the real pipeline lives, so repeat the test with a second rep and a third.

In inventory, count goods receipts that cite a purchase order against those keyed as loose entries with no order behind them, week by week. Loose receipts mean the store clerk skips the order, so price checks and three way matching never happen, and the invoice gets paid on trust. A reorder level that never triggers a requisition is another tell, because the buyer is working from memory and the shelf rather than from the system.

In a service or repair workshop, measure the job cards closed on the same day the work finished, then compare that count with the cards opened that week. Cards closed in a batch on Saturday mean the technician writes on a paper docket first, then a clerk keys it in. Watch parts consumed against the job card too, since missing spare lines distort costing and warranty claims long after the repair is done.

Each needs a saved report or filter. No survey, no meeting, just the register.

Compare the day a record was created with the day of the work, using the created-on stamp and the date the user typed in the field. A steady gap of several days is back dating, and it tells you people record work from memory or from a notebook.

Next, count blank optional fields such as mobile number, lead source, batch number, discount reason and delivery slot, and rank them by how often they stay empty. An empty field usually has a cause. It may be useless, misplaced, or buried three taps deep on a handset that the user holds in one hand while standing at a loading bay.

Treat a high blank rate as a design question first and a discipline problem second, and ask the user before you write any warning email.

Put the same five measures side by side for every branch or team, on one sheet, with the same rows for each, and keep it dull. One branch posting receipts against orders while another does not tells you more than any overall average, because it points to a person, a habit or a missing screen.

Visit the weak branch before drawing conclusions. Sit at the counter for an hour and watch the exact moment staff leave the system for paper, a phone or a spreadsheet. Often a slow ledger report, a missing barcode scanner or an unreadable tablet layout explains it, and the strong branch has a workaround worth copying.

Start small this month. Pick one measure, either a next step on every open deal or receipts against orders, then build the report, share it weekly and review it with branch heads.