How to Estimate More Accurately
Steve Barsh
April 8, 2021
  |  

How to Estimate More Accurately

Steve Barsh
April 8, 2021
  |  
Company Forensics - Learn from the mistakes of VC-funded startups | Product Hunt

Join 100,000 entrepreneurs who read us every month

We've got it!
You'll be hearing from us soon.
Oops! Something went wrong while submitting the form.

A critical skill that founders and startups need is estimating accurately. But, this is a huge area where most people struggle. Think of all the estimates you need nearly every single day as a founder running a startup:


-When will the next product version be out?

-What do our quarterly and yearly sales forecasts look like?

-How many people should we hire and when?

-How much should we raise?


The list of what you need to estimate is nearly endless. And if you can’t *accurately* estimate you are going to be way off in so many things.  And NEWS FLASH...: you will almost always UNDERestimate (be overly optimistic) and not OVERestimate. Getting estimates wrong can have devastating consequences to you, your startup and your credibility!

So, how do you fix this? How do you dramatically increase your ability to consistently deliver more accurate estimates?


Let me introduce you to your new friend:

Confidence-Based Estimating or CBE. A straightforward technique that’s easy to put into action.

With confidence-based estimating, whenever you make an estimate, I want you to get into the habit of thinking about the *confidence level* (expressed as a percentage) of your estimate.

Before I go further, let me give you a concrete example of what I mean by “confidence level” and the psychological impact. So, let’s talk about -- the weather!

If I told you that my estimate of the weather tomorrow (let’s call it a “weather forecast”) is that it's going to rain, you would think “okay, it’s going to rain.” But then if I told you “hey, there’s a 50% chance of rain tomorrow (also called the probability of precipitation or “POP”) - you may think very differently about my forecast. And if I told you there was a 25-30% chance of rain tomorrow - well that’s a completely different kettle of fish. I just went from “it’s going to rain tomorrow” to “there’s a 25-30% chance of rain” and now, you would probably leave your house and not even bring an umbrella nor rain jacket. So you not only heard the forecast, but also my level of confidence, which has a very strong influence on what you think and do.

So let’s go from talking about the weather back to talking about startups.

Here’s what I want you to try. Next time you are working with e.g, your product engineering team and they give you an estimate of when something will be done (e.g. We think we’ll get the next release out in 2 weeks) immediately ask them “What’s your level of confidence, on a scale of 0-100% that you’ll fully deliver that functionality in 2 weeks?” And figure 95%+ is like you are betting your paycheck on it. 

Wait for the slight pause… and then most likely you’ll hear a confidence level of around 60-80%. Take a breath… then I want you to follow that answer with… “okay. So what’s your forecast with a 95% level of confidence?” and wait for that answer. But brace yourself. Get ready for the next answer to go UP by 50-100%!

And, there is a chance someone will say “I can’t give you a higher level of confidence. I can’t be that sure.” That’s okay. Just realize that you need to take their estimate with a grain of salt as there is a good chance it’s not going to be accurate.

Think about that… When you first asked “when will it be done?” you meant “done.” You did not ask for an estimate at a 70 or 80% level of confidence. And now when you ask “for realsies” with a high level of confidence, “when will it be done?” you get an estimate that’s usually MUCH, much bigger.

I want you to pause and think for a moment about this >> Have you ever noticed that most things often take twice as long and cost twice as much and everyone is surprised? Ya know why? Because most people are VERY poor at estimating and no one is challenging them to think critically about their estimate. So really, should anyone truly be surprised?


Okay… so let’s keep going on this CBE thinking and and hit another level of detail that you can add to this estimation process.

As you ask for more accurate estimates with higher levels of confidence, follow up “at a 95% level of confidence” question with one more: “what are the top assumptions you are making for your new estimate?”

Let’s talk through an example. In a startup where you are building cloud software, you could hear people state assumptions like “well at a 95% level of confidence, our top assumptions that we can get this done in 2 weeks are…”


-“The new outsourced design team delivers on time.”

- “The testing team with 2 new members gets the tests done on time.

-“And, the dev team doesn’t have to do any firefighting of problems and can focus solely on working on the new release.”

Now, take those assumptions and walk through them. Talk about if they are realistic or not. Each assumption represents a risk to delivering on time. Don’t be afraid to challenge assumptions if you think they may be incorrect and possibly impact the accuracy of the estimate. For instance, from the above, YOU are fully expecting bugs to pop up and the dev team may need to take up to 25% of their time doing emergency fixes. So you should talk through that assumption, figure out how you can best manage it, and understand the realistic impact to the estimate.

Before I add one more refinement to the CBE process, I want to revisit a few great places to use the technique. Think about how you can improve your ability to estimate around questions you’re challenged with frequently:

-When is the next release going to be done?

-What is our sales forecast for next month/quarter/year?

-How many people should we hire and when?

-How much money should we raise?

-When will that customer purchase?

-When will we hit our goals?

Every single one of these is an estimate where taking a CBE approach will yield huge dividends.


Now a few more refinements to the technique and thinking:

When estimating, try to estimate “in the “small” and not “in the big.” Instead of estimating an entire project, break it up into its smaller parts and “estimate in the small.”

When you “estimate in the small” you may also catch more granular assumptions you may have missed if you were only thinking about the project as a whole.

As you bring the small estimates together and roll them up, don’t be tempted to think “that can’t be right. It can’t take nearly long /cost that much!” Stick to the number.

Next refinement -- try to do team estimates where you each team member creates estimates individually. Then bring the team together to discuss each estimate including level of confidence and key assumptions. That way you can triangulate and get a more refined, accurate estimate. By the way, be careful when the team comes together. Often very divergent estimates and opinions will come out so don’t let people “bully” others about their estimates. Stay calm. Stay critical. And go for accuracy.

So that’s it! That’s confidence-based estimating. Super simple…:

-Once you get an estimate, ask about the confidence level and push for an estimate at a 95% level of confidence.

-As you get estimates with higher levels of confidence, ask what top assumptions are being made.

-Try to create estimates at the smallest level -- remember, estimate in the small, not in the big.

-If working across a team, have each team member do their estimates independently and then bring everyone together to discuss.

Do that and your estimating accuracy will go WAY up!

By the way… one last point about estimating more accurately. Not only will your projects and company run more smoothly, but in the eyes of investors, you’ll look like a rockstar and have greater credibility. You’ll be known as an entrepreneur who delivers on what they say and when.