Every business that buys a CRM has the same conversation about nine months later. The pipeline in the system doesn’t match the pipeline in anyone’s head. Deals that closed are still sitting in “proposal sent”. Leads nobody has spoken to since March look identical to leads from last Tuesday. Someone suggests a training refresher.
The training refresher doesn’t work, because the problem was never that people didn’t understand the software.
The actual arithmetic
A salesperson finishes a meeting at a customer’s warehouse. They have four pieces of new information: the contact’s name, roughly what the customer wants, a price expectation, and a promise to follow up on Thursday.
Recording that properly costs them: unlocking a laptop or finding the app, logging in, searching for whether this company already exists, creating a contact, creating a lead, choosing a stage from a dropdown that doesn’t quite describe the situation, typing a note, and setting a task with a date. Call it six minutes and a fair amount of attention, none of which they have while standing in a car park.
Not recording it costs them: nothing today. They’ll remember. And mostly, they do remember — for about two days.
The reason CRMs don’t get updated is that the cost of recording is front-loaded onto the individual, while the benefit is diffuse and lands on somebody else. The rep pays six minutes now; the sales manager gets a better forecast next month. That’s a bad trade from where the rep is standing, and no amount of process enforcement changes the underlying maths.
Why the usual fixes don’t help
Three things get tried, in roughly this order.
Simplify the form. This helps a little, and it’s worth doing. But the remaining cost isn’t really the number of fields — it’s the context switch. Stopping what you’re doing, getting to a screen, and operating an interface is most of the expense. Removing three fields from an eleven-field form doesn’t change that.
Mandate it. Make CRM updates a condition of something people care about, like commission. This produces data, but it produces it in the shape of the mandate: a batch of records created on Friday afternoon, reconstructed from memory, technically complete and substantively thin. You’ve bought compliance rather than information.
Hire someone to do the data entry. Now the information passes through a person who wasn’t in the room, which loses exactly the detail that made it worth recording. It also costs more than the software.
What actually shifts the balance
The only durable fix is to make recording cheaper than remembering. Not slightly cheaper — obviously cheaper, at the moment the information exists.
That means the interface has to be something the person is already holding, already unlocked, and already using for this customer. For most sales teams in India, that’s WhatsApp. And the input has to be something you can produce while walking, which means speech rather than typing.
The difference isn’t marginal. “Add a lead, Kumar from Kumar Textiles, 98765 43210, wants two LED panels, follow up Thursday” is an eight-second voice note. It’s faster than deciding to remember it later. Once that’s true, the behaviour changes on its own, and it changes without a mandate — which is the only kind of change that survives the quarter.
The part that’s easy to get wrong
Making capture cheap creates a new risk: it becomes easy to write bad records. Dictation mishears digits. Company names sound like other company names. A system that silently writes whatever it thought it heard will quietly fill your CRM with plausible-looking wrong data, which is worse than an empty CRM because you’ll trust it.
So the confirmation step matters more than the capture step. The assistant should read back what it understood — the name, the number, the product — and write nothing until the person agrees. That single round trip costs a couple of seconds and is the difference between a CRM you can rely on and a CRM full of confident errors.
It should also refuse to do arithmetic. If a quotation’s prices and tax come from the assistant rather than from your own system, you now have two sources of truth for a document a customer might dispute. The assistant’s job is to record the instruction; your CRM’s job is to compute the numbers.
Where this leaves you
If your pipeline data is stale, don’t start with training. Start by working out what it costs one of your reps, in seconds and attention, to record a single lead at the moment they learn about it. If the honest answer is more than about fifteen seconds, that’s your adoption problem, and no process will fix it.
MrKattappa exists because of that number. He lives in WhatsApp, understands voice notes in any language, powered by Google Gemini, confirms what he understood, and writes the record into FunFlow CRM — under your name, with your permissions. See what he handles, or put him on the payroll for ₹1,000 per user per month.
Chat on WhatsApp