Choosing a software company in Bangladesh comes down to five checks: whether the team has built something close to what you actually need, who owns the code and the intellectual property when the work is done, how the engagement is structured and priced, who you will be speaking to when something goes wrong at 2am, and what independent evidence exists for the claims on the website. This guide sets out what to look for under each, the questions that separate a serious supplier from a plausible one, and how to verify the answers before you sign anything.
Disclosure: this guide is published by Invento Software Limited, which is itself a software company in Bangladesh. We have written it as the brief we would want a buyer to hold us to. We describe our own work in a clearly marked section at the end rather than ranking ourselves against other firms.
The software industry in Bangladesh today
Bangladesh exports software worth around $1.6 billion a year to roughly 80 countries, and there are more than 2,600 registered IT companies listed with BASIS, the industry association. The majority offer software development services of some kind.
That scale is the opportunity and the problem at once. There is genuine engineering depth in Dhaka, and rates sit well below Western markets. There is also a long tail of firms whose websites are indistinguishable from one another, because the same service list, the same technology logos and the same promises appear on nearly all of them. Nothing on a homepage will tell you which is which. The checks below are what will.
Two practical points that catch first-time buyers out. The standard working week in Bangladesh runs Sunday to Thursday, so a Friday deadline agreed with a European or American office lands on a weekend in Dhaka. And the country sits at GMT+6, which gives comfortable morning overlap with Europe and very little with North America. Neither is a problem, but both need to be designed into the working arrangement rather than discovered in week three.
What to judge a software company on
Most vendor shortlists are built on the wrong signals – office photographs, client logos, and the length of the technology list. The six criteria below are the ones that predict whether a project actually lands, and each has a way of being checked rather than taken on trust.
1. Relevant delivery, not the client logo wall
A logo tells you a company was paid by a recognisable brand. It does not tell you what was built, how big the team was, or whether the project resembles yours. Ask for one project in your industry and one at your size, and ask what went wrong on each and how it was handled. A supplier who cannot name a single difficulty on any past project is either not being straight with you or was never close enough to the delivery to know.
2. Ownership of code and intellectual property
This is the single most expensive thing to get wrong, and it is settled in the contract rather than the pitch. Establish in writing who owns the source code, what happens to it if the relationship ends, whether any third-party or in-house framework is embedded in your product, and what licence governs it. A firm building on an open-source platform such as ERPNext or Odoo should be able to explain exactly which parts of your system are the platform, which are custom, and what that means for upgrades later.
3. How the engagement is structured
Fixed-scope, dedicated team and staff augmentation are different commercial instruments, and picking the wrong one causes more failed projects than poor engineering does. Fixed-scope suits work you can specify completely in advance. A dedicated team suits an evolving product where you set the priorities. Augmentation suits a known roadmap and a missing skill. If a supplier pushes one model regardless of what you describe, that tells you something.
4. Who is actually assigned
Ask who will work on your project by name and seniority, and whether those people are currently on another engagement. Ask what happens if one of them leaves. In a market where good engineers move, continuity of the specific people matters more to your delivery than the total headcount of the company – and headcount is the number most often quoted.
5. Domain understanding
A team that has already built for your sector will argue with your specification, which is what you want. One that has not will implement it exactly as written, including the parts that are wrong. For regulated or process-heavy work – manufacturing, distribution, healthcare, finance – domain familiarity reduces the cost of the specification far more than a lower hourly rate does.
6. Stability of the team
Attrition is the quiet risk in outsourced development. Ask directly what the company’s turnover has been over the last two years and how long the average engineer has stayed. A supplier that tracks this and answers plainly is telling you something useful. One that has never measured it is also telling you something.
Questions worth asking before you shortlist
- Which project in your portfolio is closest to ours, and why?
- What went wrong on it, and what did you change afterwards?
- Who specifically would be assigned, and what else are they working on?
- Who owns the code, and what is written in the contract about it?
- What is not included in this price?
- What happens after launch – who fixes defects, and for how long?
- Can we speak to a client you worked with two years ago, rather than last month?
That last one is the most revealing question on the list. Recent clients are still in the honeymoon period. A client from two years ago knows whether the support held up.
How engagements are usually structured
Most work in this market falls into one of three shapes. A fixed-scope custom software project works when requirements are stable and can be written down in full. A dedicated team works when the product is still evolving and you want to set priorities sprint by sprint. MVP development is a narrower case of the same thing, aimed at getting a testable product in front of users before committing to the full build.
Specialist work – QA and testing, cloud engineering and DevOps, user experience design, mobile app development and website design and development – is usually bought either as part of one of those arrangements or as a defined piece alongside your own team.
Warning signs
- A quote arrives before anyone has asked what you are trying to achieve.
- The proposal is priced but the scope is described in a paragraph.
- Performance figures appear on the website with no source, no period and no definition – a “success rate” means nothing unless you know what was counted.
- Client references are offered but never actually arranged.
- Every technology you mention is met with “yes, we do that”.
- Nobody technical is in the sales conversation.
How to verify what you are told
Check BASIS membership against the association’s own member list rather than the badge on the website. Ask for the company registration details and confirm the trading history. Ask for two references and actually call them, with your own questions rather than the ones you are handed. Where a public case study exists, ask the supplier to walk you through the parts that are not in it. And put the code ownership, the handover terms and the post-launch support period into the contract before the first invoice, not after the last one.
About Invento Software Limited
Invento Software Limited publishes this guide. We started in 2014 with three people, initially in web design, and moved into enterprise software – principally ERPNext implementation, customization and open-source enterprise systems – alongside custom development, mobile applications, MVP work, QA, cloud engineering and design.
- Founded: 2014
- Location: House 484, Road 32, Mohakhali DOHS, 5th floor, Dhaka 1206
- Working days: Sunday to Thursday
- Clients include: Radisson, Sundora, United Fortune Inc, Envoy Textiles, Pan Pacific Sonargaon, Syngenta, HSBC, Tamishna Group, Forever, Haptex and Square Textiles Limited
- Contact: hello@invento.com.bd
We would rather be assessed against the criteria above than against a ranking we wrote ourselves. If you are running a selection process, ask us the questions in this guide – including the awkward ones about what went wrong and who owns the code.
Choosing well
The decision that matters is not which company is best in the abstract, but which one is right for the specific thing you are building, at the size you are building it, under terms you have read. A supplier who welcomes hard questions about ownership, staffing and failure is usually a better bet than one with a longer client list and vaguer answers. Verify the claims, get the terms in writing, and speak to a client who has lived with the work for two years.
