Cover is worked out, not guessed
Replacement options are calculated against travel time, availability and contract requirements. The system proposes a small set of workable choices; the manager decides.
Case study — work in progress
Manage by Text: catching the no-show before the client does
A WhatsApp-native operations system for commercial cleaning firms — attendance, replacement cover and proof of service, with no app for the workforce to install. This build is in progress; what follows is the problem, the design and the decisions made so far.
Status: work in progress
Status: work in progress
This product is being designed and built. There are no client results to report and no live version to point at yet. We are showing it because the thinking is the interesting part, and because a portfolio of finished work alone would be a misleading sample.
The problem
Contracts die from unmanaged no-shows
Commercial cleaning runs on thin margins, and the lever that matters most is not efficiency — it is retention. A contract is lost when a site is left uncovered, the client notices, and confidence goes. The workforce turns over heavily, so the person who was supposed to arrive is often new, unavailable, or simply not answering.
Incumbent tools assume staff will install an app, remember a login and clock in. For a dispersed, high-turnover team, that assumption is where the system dies. Meanwhile the operation already runs on WhatsApp group chats: the shift list, the swaps, the “who is covering this?” messages.
The intended loop
From attendance truth to proof of service
-
1. Clock in by message
A cleaner sends ARRIVED at the site and FINISHED when done. No app, no new account — the same WhatsApp thread the team already uses.
-
2. A no-show is a fact, not a phone call
When a site is not confirmed, the system raises it immediately rather than waiting for the client to complain.
-
3. Cover options, worked out
The system calculates a small set of replacement options, each balancing a different priority: least disruption, lowest cost or fastest arrival.
-
4. The manager chooses
The manager picks an option. The system sends the minimal set of changed instructions to the people affected, rather than redistributing the whole day.
-
5. Proof reaches the client
Attendance, photos and completion records become a service record the cleaning firm can show, not a claim it has to make.
Decisions
Where the intelligence is allowed to sit
The chat is the surface, the record is the truth
Messages carry the conversation; attendance, assignments and the audit trail live in a proper record. The group chat is where people work, not where the business is stored.
Its own product, not a template
It shares a product family and hard-won lessons with Trade by Text, not a design. A sole trader's quote-to-cash flow and a cleaning firm's shift operations are different problems.
Proof, not promises
Attendance, response to cover requests and completion records build a service history the firm can show a client, rather than a claim it has to make.
Where it is
Started from a real operation
The work began with a real Dorset cleaning firm, not a hypothetical: mixed residential and commercial work, holiday-park contracts, and staff moving between sites during the day. The design covers the attendance model, the cover interaction and the rules for who can be offered which shift.
Discovery, the specification and a phased build plan are written, and the build is following them. There is no live client installation and no results to report yet.
Follow the build
Follow the build
managebytext.com is where the product will live. In the meantime, the closest live example of the WhatsApp-first approach is Trade by Text.
- managebytext.comThe product site, still in progress.
- Trade by TextThe same philosophy, live, for tradespeople.
- DiscoveryHow we scope a system like this.
Questions
The things people ask first
Is Manage by Text live?
Not yet. It is in design and build. We are publishing it as a work in progress rather than waiting for a launch, and we will not present results until there are real ones.
Why not build an app for cleaners?
Because a high-turnover workforce will not reliably install one. WhatsApp is already installed and already used for the shift conversation. Removing that barrier is the point of the product.
Does an AI decide who covers a shift?
No. Cover is worked out from travel time, availability and contract requirements, and the manager makes the final call. A model may help understand a messy reply or draft an explanation, but it does not make the assignment.
Is it the same product as Trade by Text?
No. They share a name pattern and some hard-won lessons, not the same product. Trade by Text handles quoting and payment for sole traders; Manage by Text handles attendance and cover for cleaning firms.
Related
Let’s find out whether automation is the answer
Tell us what is slow, expensive or error-prone. We will tell you honestly whether we can help, what it would take, and roughly what it would cost — before anyone signs anything.