How to Select an Optimization Company

Getting started with any sort of optimization can be daunting; you don’t want to disrupt your operations, and you need to make sure things work smoothly before implementing anything.  You’re already busy with day-to-day operations.  Here’s our guide to making the process as quick and smooth as possible.

1. Identify your specific needs

The first step in the process is identifying your needs.  You probably already have some idea of which aspects of your operation you’d like to improve; it might eventually be several things.  

In this stage, however, you should try to be as specific as possible - separate all the different functions that you would like to automate or optimize.  

Try to label each of the specific needs with as many criteria as you can - one example would be labeling if the need is operations-critical or not (ie. do we absolutely NEED this to work all the time, or can we fix it as we go).

Being as specific as possible here is really important.  Make sure you define the terms you use, and break things down if they have sub-categories.  One example would be within ‘crew scheduling’, you might have ‘cabin crew days off’, ‘cabin crew training’, ‘pilot days off’, ‘pilot training’, ‘maintenance days off’, etc.

2. Prioritize these needs

Once you have your full list of needs, it’s time to prioritize.  This step will require some decisions on your part, as you’ll need to pick the criteria by which you prioritize.

Some of the common criteria are pain, time, and cost.  

Pain is a bit of a qualitative criterion, and you use something like a 1-10 scale to rank needs.  It’s often a representation of time, but it encompasses a little more of how painful the task is to complete, how much your current team dislikes it, how far it is from your core business or mission, etc.

Time is fairly obvious, though you should really make sure you consult each department to make sure you have an accurate picture of how much time really is spent on each step.  We’ve seen examples where departments were spending more than 80 hours per week (as a department) on something that other departments assumed was trivial.

Cost can be measured in a couple different ways. It can be another proxy for time - the cost of the time that employees spend on a given task - or it can be more direct: an expected reduction in the number of empty miles or hours that could be provided by optimization.  

It could be a combination of both (ex: automating and optimizing aircraft assignment to booked routes would save both time scheduling, and reduce the amount of empty distance flown).

The other main thing you should identify here are the projects you perceive as relatively small, but would have a disproportionate impact.  Some examples might be tasks that are highly manual, but likely easy to automate.  You may need to consult with an optimization provider to confirm which projects these might be, but do it yourself and revise later.

3. Tentatively pick a small starting project

Based on the previous step, you should now have a ranked list of projects, perhaps with some with a star next to them to identify those you believe might be small projects with a disproportionate impact.

In our experience, a lot of the failed attempts to work with optimization companies can be traced back to tackling too large a project up front.  When this happens, both the optimization company and the customer end up frustrated, because things aren’t moving quickly enough, the results they’re seeing are sub-optimal, and the provider is underpaid for the overall effort they’re putting in - a bad situation for everyone.

In this step, you should pick one of the projects you identified that you believe would be small, but have a disproportionate positive impact on your operations, and plan to seek someone to complete this for you.

At this point, we should mention again that all the previous steps have been completed under the “magic wand” mentality - this assumes the ideal scenario where everything is possible.

4. Identify your list of optimization companies

At this point, you need to start seeking vendors.  You’ve likely done this before, so we won’t spend too much time on it, though for optimization companies it’s likely a little different.  Some of the best possible partners may not be currently in your industry, or may operate in a slightly different industry (commercial aviation vs. business aviation, for example).

Potentially good sources of leads:

  • From within your own team

  • Industry partners and friends from other companies

  • Tradeshows

  • Crunchbase, AngelList

A real test of an optimization provider's confidence is to get them to recommend a couple other companies to check out.  This is actually really important, as depending on who they recommend, you can learn a lot.  Some examples:

  • If they recommend some providers known to be good, you’ll know they likely value their reputation and are solid themselves;

  • If they recommend providers known within the industry to be poor, you’ll know they don’t know the industry well, or have low standards;

  • If they refuse to recommend anyone, won’t disclose why, or give a poor reason, you can infer they aren’t very confident in their technology.

5. Rank each company identified based on fit for your problem.

To complete this step, you should try and gather as much information you can about each company - read their blogs, download white papers, etc., to try and get a better picture of the types of problems they tackle and the abilities of their technology.

Now, ranking based on ‘fit for your problem’ can be confusing - fit for your small starter problem, or fit for your whole vision?  

Our suggestion is rank for both, or make separate lists if you want to.  Our reasoning for this is often a small company that could tackle your small problem more effectively might not have the capability to fulfill your whole vision, while a big company that could fulfill your whole vision might not be flexible enough to solve just the small problem without imposing a much larger product than you’re comfortable starting with.  Ideally, you want the best potential candidates for each.

6. Reach out to the top few in each case, and request calls/demos.

When you reach out, you should frame your complete vision, and the first need (or few needs) you’ve identified to solve, and mention you’d like to have a call to chat in more detail to find out if the provider might be a fit.

While there is some value in demos, and some companies will want to do their standard demo for you, we’d argue you want more of an interactive call; one where you describe in more detail your needs, and then they can describe their technology and how it might help.  We think it’s a much better use of time than the stock demo.

7. Narrow your list to maximum 6 companies, and bring in a third party

Maximum 6 means maximum 3 big and 3 small, and if you can, make it less.  The main point of this step is to bring in a third party who you feel comfortable with evaluating the technology.  

The aim here is to get an outside perspective, and also to actually be able to evaluate the underlying technology; make no mistake, this is going to be difficult.  

The people who can truly understand the strengths and limitations of various types of optimization technology are few and far between, but if you don’t complete this step, you’ll be relying on trust in your contact at the provider, or the referrals of others.

An ideal third party would be someone who has used optimization technology before, has a background in operations research and/or computer science, and has worked in logistics, preferably in the same, or a related branch, as you.

8. Schedule a more detailed follow up call with the third party and prospective providers

This is essentially due diligence - you don’t even have to be part of the call if you don’t want, though we’d recommend it.  The point here should be allowing the third party to get a handle on how each company operates, and the details of the technology behind their products.

Once you've shared the details you compiled earlier around your needs and priorities, they should then be able to make a recommendation to you.

9. Pick one company, and engage with them

This whole process, while it probably seems long through this post, shouldn’t take long (<2 months).  

It’s extremely frustrating for all involved if the evaluation and engagement process takes months on end.  Providers will feel like you’re using them, and lose faith in the deal actually closing.  

You should provide a timeline for providers on the next expected steps, and when they will be completed, and stick to those steps.  It will provide a solid foundation of trust for when you engage with them, and they’re expected to stick to a timeline.  If you won’t do it for them, why should they do it for you?

Once you’ve picked a company, sign an LOI (providers love this, and they’re not legally binding for you).

At this stage, you can enter the final phase of evaluation and ask for some reference customers from the provider.  They should provide you with one or two that are willing to chat with you, and you can follow up with them to verify that they’re happy with the products and services that have been provided.

10. Develop a trial and full-contract agreement, and get started

At this point, you need to figure out the details of the deal.  We’d recommend signing a 3-5 month trial agreement, as well as a contract for following the trial (as an option).  That will give you an idea of the costs associated once the trial is done, and also limit your downside if the trial doesn’t work.  

Providers will want this information as well so they can plan effectively.

Expect the trial agreement to be more expensive than the ongoing cost of the software, unless you plan on paying for extra levels of support; in the beginning, you want to have access to lots of support, and have a great setup experience, and you should view this as consulting time and an investment. 

If it’s a software-as-a-service company, settle on terms for the ongoing monthly payment.  Fixed or tied to a particular metric (flight hours, crew members, etc.) is up to you, but you should settle on terms before starting the trial so the post-trial transition is smooth.

Again, you should aim to complete this step as quickly as possible.  Building mutual trust on timelines, and being prompt, will pay dividends down the line as you continue to work with each other.

Get a PDF version of this blog post: