Custom CRM or ERP in Kazakhstan: will automation pay off?

Comparing “a subscription to a ready-made CRM versus building your own” answers which is cheaper to own. But the decision to automate is driven by effect, not cost: the system must return more than it costs. That effect can’t be taken from a price list — it has to be measured in your process. So a payback calculation starts not with a contractor’s proposal but with a baseline of how the work is done today.

The horizon comes from the business: how far ahead you can plan the process without radical change. If the answer is “six months” and payback lands later, that is a signal to start with a smaller first phase or a ready-made product.

Without a baseline, a year later nobody can say whether the system had an effect. Pick a typical period without seasonal peaks and collect data on the operations the system is supposed to change. Use what is already recorded: 1C exports, messenger history, call logs, spreadsheets.

Record the spread, not only averages: if the same operation takes very different time for different people, the problem may be training or procedure, and that is cheaper to fix without development.

The effect has several components. Calculate each separately and write down the assumptions — those are what will be checked later.

• Time: E_time = Σ (t_before − t_after) × N × C_hour, where t is operation time before and after, N is operations per month and C_hour is the fully loaded hourly cost of an employee.
• Errors: E_errors = (n_before − n_after) × C_fix, where n is errors per month and C_fix is the time and direct loss per error.
• Lost enquiries: E_leads = ΔN_handled × K × A × M, where K is the share of enquiries that become deals, A is the average order value and M is the margin.
• Turnover: the effect of invoicing, shipping or reconciling faster — through the cost of the cash or stock tied up for that time.

Run three scenarios — cautious, base and optimistic — and decide on the cautious one. If the project pays off only in the optimistic case, it is a bet, not a calculation. For each scenario, record which values of t_after and n_after you assumed and why.

The one-off investment I and monthly running costs C_month must include everything the system will need over the horizon, not just building the first version.

For a ready-made service the same formula looks different: a subscription for the right number of users including team growth, integrator work and — the part most often forgotten — staff time spent on workarounds where the process doesn’t fit the system. A detailed cost-of-ownership comparison of custom and ready-made CRM is in separate blog articles linked at the end.

These are the factors that change integration scope and hosting requirements in local projects. Find them out before any estimate: each can shift the one-off investment noticeably.

• The process hasn’t settled and changes every few weeks: the system would lock in what becomes obsolete tomorrow.
• Volume is small, and even the optimistic effect doesn’t cover the monthly running costs.
• A ready-made service covers the process with reasonable configuration, and there are almost no workarounds.
• There is no internal owner to make decisions about the system and look after data quality.
• The effect is described as “more convenient” but can’t be expressed in time, errors or money.
• The source data is in such a state that migrating it would be a project of its own.

In these cases it is wiser to configure a ready-made system, automate one narrow area — for example, only the order exchange between the website and 1C — or first tidy up procedures and data. The calculation above still helps: it shows where the effect is greatest.

The checkpoint is what separates a calculation from a forecast. If the actual effect falls below the cautious scenario, revisit the next phase before money goes into it.