How to buy technology without getting burned

A wrong purchase leaves a tool that does not fit, a contract you cannot leave and years of cost. Here is how to buy well.

Five common mistakes

Mistake	Do this instead
Buying on the demo	Run a trial or proof of concept with your own data, tested by the people who will use it.
Letting the vendor set the requirements	Write down the problem, what success looks like, and must-haves against nice-to-haves first.
Ignoring total cost	Model every cost over three to five years. A cheap tool can cost two to three times its price.
Skipping references	Call customers like you in size, industry and use, about rollout, support and results.
Buying too much	Choose the tool that meets 80% of needs at half the price over the premium one.

Our work with a financial services firm showed how much waste piles up when companies overbuy.

A five-step evaluation

1. Define the requirements

• Business. What problem are you solving, and what outcome do you need?
• Function. What must the product do, feature by feature?
• Technical. What must it connect to, and which security, compliance and performance rules apply?
• Operations. What support hours, SLA, training and admin effort do you need?
• Budget. What can you spend, including setup and running costs?

2. Scan the market

Use analyst reports, peers, review sites and industry forums to build a long list of six to ten options.

3. Shortlist

Score each against your requirements and keep three or four. Skip anything that clearly does not fit.

4. Evaluate the finalists

• Ask for a demo built on your use cases.
• Run a proof of concept for a large purchase.
• Check references from similar organisations.
• Review the vendor’s finances and product roadmap.
• Review the rollout plan, its timeline, its people and its risks.
• Read the contract before you grow attached to a choice.

5. Negotiate and decide

Negotiate with your top two at the same time, since a real alternative gives you leverage. Decide on total value, not price alone.

Negotiating the contract

• Know the pricing model. Ask what you would pay at 1.5 and 2 times your current size.
• Choose the term. Multi-year discounts lock you in. If you doubt the fit in three years, take a shorter term with renewal options.
• Write the exit. Set termination terms, data export, a transition period and any early exit fee. A vendor that resists an exit clause is telling you something.
• Cap price rises. On a multi-year deal, cap yearly increases at 3% to 5%.
• Get commitments in writing. Promises about features or rollout time vanish once the contract is signed.

On the SLA, negotiate these points.

• Uptime guarantees with real financial penalties.
• Response times by severity.
• Resolution targets as well as response times.
• Escalation to named people, not a ticket queue.
• Reports that let you check the SLA was met.

Hidden costs

• Setup and customisation can equal or exceed the first year’s licence fees.
• Integration may need custom code, middleware or extra tools, so get quotes first.
• Training costs fees, a dip in productivity and ongoing training for new hires.
• Data migration means cleaning, mapping, checking and testing, which is never as simple as promised.
• Administration may need a dedicated person or regular upkeep.
• Overage charges apply above plan limits on usage pricing.
• Add-ons can price features most customers need as upgrades.

Build or buy

Buy when a proven product meets 80% or more of your needs and the capability does not set you apart. Buying also suits you when speed matters, or you lack developers to build and maintain software.

Build when your needs are truly unique or the capability sets you apart. It also fits when you have the developers, or when buying would need heavy customisation. Data rules that exclude third parties point the same way.

Often the best answer mixes both. Buy the standard capabilities, and build only what sets you apart.

Managing licences

• Audit quarterly. Most organisations find 15% to 25% waste in unused licences, duplicates and silent auto-renewals.
• Centralise buying. Departmental buying brings overlap and missed volume discounts, so require approval for new subscriptions.
• Watch usage. If you pay for 100 licences and 60 people log in, you are overpaying.
• Plan for true-ups. Some enterprise licences charge each year for users above your count.
• Consider a software asset management tool. It finds, tracks and flags savings, and often pays for itself in the first year.

An IT strategy advisor who earns no commission can build a buying process that saves money and cuts risk.

Questions people ask

Are we paying too much for software?

Compare with similar companies through public price pages, peers and vendors’ own data. If you are well above the benchmark, renegotiate at renewal. Check usage too, since you may pay for licences nobody uses.

Should we use a procurement consultant?

Under $50,000, most companies can buy well with a structured process. For larger, complex purchases such as ERP or infrastructure, an independent consultant usually pays for itself. Choose one who earns no commission on what they recommend.

What is the biggest mistake?

Talking to vendors before defining requirements. You end up judging products on their strengths, not your needs. Features you will never use then justify a higher price.

What if a vendor will not negotiate?

Get competing proposals, and negotiate at the end of a quarter or fiscal year. Offer a longer term for a better price. A vendor that will not move on anything is showing how it will treat you as a customer.

How early should we plan a big purchase?

For $50,000 or more, start six to nine months ahead. That leaves time to define needs, compare options, negotiate from strength and plan the rollout.