CRM for a construction company: from lead to project handover
A construction company sells not a product but a promise: that the project will be delivered on time, within budget, and without surprises. Between the client's first call and the final signed act there aren't three days as in an online store — there are six to eighteen months, dozens of people, hundreds of material line items, and several revisions of the estimate. Excel holds up exactly until the second project. Then the familiar story begins: the foreman can't remember how many bricks were delivered to Lisova Street, the client is sure the balcony was included in the price, and the accountant can't work out whether the company earned anything at all. Let's look at what a CRM for a construction company should actually be so this doesn't happen.
Why a generic CRM for a construction company breaks
A generic CRM is designed for a short cycle: the lead comes in, the manager calls, and within three days the deal is either won or lost. In construction the sales cycle is measured in months, not days. A client asks “how much does a 120-square-meter house cost,” gets a rough estimate, disappears for two months, comes back with a new floor plan, negotiates, waits on a bank's mortgage decision — and only then signs the contract. A standard five-stage funnel turns into a graveyard here: half the deals hang in “thinking about it” for months, the manager has no idea who to call today, and the owner stares at a sales forecast that bears no relation to reality.
The second breakdown is more serious. In a generic CRM the deal closes the moment the contract is signed — and there the system falls silent. In construction, that's exactly where everything begins: the estimate, purchasing, crews, stages, acts, tranches, warranty. So the working model has two parts: a sales funnel up to the contract, and a project card after it. The project becomes the central entity to which everything is attached — the client, the contract, estimate versions, the work schedule, costs, payments, site photos, documents. One project, one folder that shows the entire story from lead to handover, and any employee can understand what's going on there in a minute.
Estimates and change orders: project tracking starts with numbers
Project tracking begins with the estimate, and the main problem isn't drafting it — it's managing changes. The client ordered laminate, then switched to parquet mid-project; added underfloor heating; moved a wall by a meter. Every such change means different materials, different man-hours, and a different price. If changes are agreed verbally, at the end of the project you get a conflict: the client is certain he's paying the original estimate, while you spent 180,000 more. The system must keep estimate versions: the baseline plus every change order with a date, scope, new total, and the client's confirmation — even if it's just a button tap in Telegram. The rule is simple: no confirmed change order, no work.
Next, the project is broken into stages: demolition, rough works, utilities, screed, plaster, finishing, handover. Each stage has planned and actual dates, an owner, and a completion act. Once a stage is closed, the system generates the act with a list of works and volumes, sends it to the client for approval, and unlocks the next tranche. This does two things at once: the client sees progress in documents rather than in words, and the company doesn't move into the next stage until the previous one is closed and paid. Projects where acts get signed “later, all at once at the end” are always the projects where the company gets underpaid and spends months arguing over volumes nobody remembers anymore.
Crews, schedules and photo reports: construction automation on site
Construction automation on site is first and foremost about allocating people. The company has three crews and seven projects; the bricklayers are needed on Lisova on Tuesday — and on Sadova on Tuesday too. Who decides? Usually the foreman, in his head, which is exactly why a crew shows up at a site where the materials haven't arrived yet and a day burns away. A schedule inside the system shows who is where and when, which stages depend on each other, and where resources collide. A four-person crew standing idle costs six to ten thousand hryvnias a day — and nobody ever sees that money in a report, because it's recorded nowhere. Over a year those invisible days add up to a month's turnover.
The simplest and most effective control tool is photo reports from the site via a Telegram bot. Every day at six p.m. the foreman gets a reminder, picks the project from a list, sends five to ten photos, and briefly writes what was done. The bot attaches the shots to the project, the stage and the date, and they land in the project card immediately. It takes two minutes and solves three problems: the owner sees the real state of the works without driving to the site; the client gets a weekly report and stops calling every day; and if a pipe leaks inside a wall a year later, you have photos showing how it was laid and who laid it. No separate app to install — everyone already has Telegram, which is precisely why people actually use it.
Project management in construction: materials, tranches, overruns
Project management in construction is above all the management of money flowing into materials. Every purchase must be tied to a specific project and a specific estimate line: not “bought 200 bags of cement,” but “Lisova, screed, 200 bags, 46,000 hryvnias, budgeted 42,000.” The system immediately flags the 4,000 overrun and demands a reason. Without that link, materials quietly migrate from one site to another, leftovers are never booked back into stock, and profit dissolves without a trace. The working norm is to keep material variance within five percent; if you're consistently at fifteen to twenty, you're no longer building — you're sponsoring.
Client payments must also live inside the project card. The classic scheme: a thirty percent advance at the start, then tranches tied to each closed stage, and five to ten percent after the final act is signed. The system holds the tranche schedule, shows how much the client has already paid, how much is outstanding, and whether you're heading into a cash gap — where the stage costs are already incurred, materials bought, the crew paid, but the tranche hasn't arrived. An automatic payment reminder three days before the due date does what a manager doesn't: it isn't embarrassed to ask, and it doesn't forget. Over time this cuts average payment delays by one to two weeks — and that is your working capital.
What a construction company CRM shows after handover
The project is handed over — and this is exactly where most companies switch the system off. That's a mistake: the warranty period runs two to five years, and claims will definitely come. Warranty requests should be handled by the same Telegram bot: the client describes the problem, attaches photos, the system links the request to the project, identifies which stage and which crew did the work, and assigns a task to the responsible person with a forty-eight-hour response deadline. If claims are instead handled in the foreman's private chats, you either lose your reputation along with referrals, or you fix at your own expense work done by a subcontractor from whom, two years on, you'll recover nothing.
And the key number all of this is built for: the profitability of each project separately. Not “the company earned X this quarter,” but “Lisova returned eighteen percent margin, Sadova three, and Vyshneva actually lost forty thousand because we redid the screed three times.” Once you see this per project, patterns emerge: a certain type of work is always unprofitable, a certain crew always blows the budget, a certain manager systematically underprices to close the deal. That's the moment a company stops growing turnover and starts growing profit. These are exactly the systems — with a project card, estimate versions, crew scheduling, a Telegram bot for photo reports, and a per-project margin report — that we build at Devlly for a specific construction company, because off-the-shelf tools break here on the very first change order.