Short answer
Every traditional business has more candidates for digitization than it has time or budget to build. The order you choose matters more than the tools you pick. The tasks worth automating first are usually the ones repeated daily, not the ones that sound most impressive to describe.
Start with frequency, not glamour
The tasks worth automating first are usually the ones repeated daily, not the ones that sound most impressive to describe. Answering the same three customer questions dozens of times a week is a better candidate for a first project than an elaborate recommendation engine that might get used occasionally. Frequency multiplies the value of even a modest improvement.
When we looked at where staff time actually went across Star Designs and Kea, the answer was rarely the exciting part of the business. It was repetitive coordination — confirming order status, chasing paperwork, re-answering questions. Those are unglamorous but high-frequency, which makes them the right place to start.
Weigh reversibility
Some digital changes are easy to undo if they do not work — a new page on a website, a revised email template. Others are hard to reverse — migrating core records to a new system, or changing how orders are tracked. I try to sequence the reversible experiments first, to learn cheaply, before committing to changes that are expensive to walk back.
This ordering also builds internal confidence. A few small, reversible wins make it easier to get buy-in from staff and partners for the larger, harder-to-reverse changes that come later.
Protect the parts that depend on judgment
Some tasks look automatable but actually depend on judgment built from years of experience — grading a stone, sensing when a client relationship needs a phone call instead of an email, matching two families whose values and expectations need careful reading. Digitizing these prematurely, or replacing them entirely with an algorithm, produces worse outcomes even if it looks efficient on paper.
The better move is usually to build tools that support that judgment — giving a grader better reference data, giving a matchmaker better background information — rather than tools that try to replace it outright.
Consider what your customers are actually asking for
It is easy to digitize what is convenient to build rather than what customers are actually requesting. We paid attention to the specific complaints and requests that came up repeatedly in customer support conversations, since those are a more honest signal of priority than internal assumptions about what would be impressive.
A recurring request for order tracking, for instance, is a clearer signal than an internal hunch that a slick new feature would delight customers. The former is data; the latter is often just enthusiasm for building something new.
Sequence for compounding value
Some digitization projects make later projects easier — clean customer records make future personalization possible; a reliable order system makes future automation trustworthy. I look for projects that lay groundwork for what comes next, rather than treating every project as an isolated bet.
This is where a list of candidate projects becomes a roadmap rather than a grab bag. The right first project is not always the highest-impact one in isolation — it is often the one that makes every subsequent project cheaper and safer to build.
Budget for maintenance, not just launch
A common mistake is digitizing a process and then under-resourcing its upkeep. Software that goes stale — outdated inventory, broken forms, unanswered chat — is often worse than the manual process it replaced, because customers now expect a response speed the manual process never promised.
Before greenlighting a new digital tool, I ask who will own keeping it accurate and responsive once the initial excitement of building it has faded. If there is no clear answer, it usually is not ready to launch yet.
What to remember
- Prioritize high-frequency, unglamorous tasks over impressive but rarely used features.
- Start with reversible changes before committing to hard-to-undo migrations.
- Support human judgment with better tools rather than trying to replace it outright.
- Digitize based on real customer requests, and plan for ongoing maintenance from day one.
Frequently asked questions
- What should you know about start with frequency, not glamour?
- The tasks worth automating first are usually the ones repeated daily, not the ones that sound most impressive to describe. Answering the same three customer questions dozens of times a week is a better candidate for a first project than an elaborate recommendation engine that might get used occasionally.
- What should you know about weigh reversibility?
- Some digital changes are easy to undo if they do not work — a new page on a website, a revised email template. Others are hard to reverse — migrating core records to a new system, or changing how orders are tracked.
- What should you know about protect the parts that depend on judgment?
- Some tasks look automatable but actually depend on judgment built from years of experience — grading a stone, sensing when a client relationship needs a phone call instead of an email, matching two families whose values and expectations need careful reading.
- What should you know about consider what your customers are actually asking for?
- It is easy to digitize what is convenient to build rather than what customers are actually requesting. We paid attention to the specific complaints and requests that came up repeatedly in customer support conversations, since those are a more honest signal of priority than internal assumptions about what would be impressive.
- What should you know about sequence for compounding value?
- Some digitization projects make later projects easier — clean customer records make future personalization possible; a reliable order system makes future automation trustworthy. I look for projects that lay groundwork for what comes next, rather than treating every project as an isolated bet.