Measuring Price Elasticity And More

Price elasticity is a concept every business person should understand but I have found that many don’t.

Wikipedia defines price elasticity as:

a measure used in economics to show the responsiveness, or elasticity, of the quantity demanded of a good or service to a change in its price

Here is a chart that, I think, makes the concept easier to understand:

In it’s simplest terms, the lower the price of something the more demand there usually is for it. But every product and service has its own elasticity curve and it is important to understand what the price elasticity is of your product or service.

The good news is that it has never been easier to determine the price elasticity curve of a product or service.

Here is how you do it.

  1. offer the product or service on the web and make the purchase as easy as possible (Stripe and/or Paypal).
  2. establish the range of pricing you want to measure, start at a number higher than you can imagine anyone paying and end at a number that is equal to the cost to produce your product or service (the cost of good sold)
  3. set the price at the high end of the range
  4. buy some search traffic to your offering (Google Adwords)
  5. measure the traffic to your offer and the conversion rate (Google Analytics)
  6. lower the price
  7. repeat 4 & 5
  8. lower the price again
  9. repeat 4 & 5
  10. continue this process until you reach the low end of the range

Then plot conversion rate against price and you will have the price elasticity of your product or service. It is best to keep everything other than price constant as you move through this exercise. For example, don’t change the adwords campaign as you move through this process.

As you do this, you can also measure what it costs to acquire a customer (CAC) via search. That may not turn out to be the best way to acquire a customer but it’s a very helpful number to know.

You will want to consider this formula as you think about where to land on pricing:

Price > CAC + COGS

That means the price you charge must be greater than the cost you must pay to acquire a customer plus the cost you must pay to make or deliver the service.

If your product or service is sold on a subscription basis, then you must also know the amount and timing of churn to expect and the lifetime value of a customer (LTV). In a subscription offering, the above formula becomes


All of these concepts and math falls under the terminology of “unit economics” and you will often hear investors (including VCs) talk about “understanding the unit economics” of a business. If you don’t know what that means when an investor brings it up, you are unlikely to close that sale.

But I am not writing this to help entrepreneurs raise money. I am writing this post to help entrepreneurs understand how to build a profitable business.

You must know the price elasticity of your product or service. You must know how much it costs to produce. You must know how much it costs to acquire a customer. And if your model is subscription, you must know your churn and lifetime value. From all of that comes the data and knowledge that allows you to optimize price, margins, and profitability. Which, after all, is the goal of a business, all the other bullshit you read on the internet notwithstanding.


Comments (Archived):

  1. pointsnfigures

    Elasticity of supply and demand are hugely important. Demand curves always slope down. Going deeper into these microeconomic concepts is pretty important. Knowing the theory and practice behind consumer/producer surplus and how it affects your business will help the startup maximize revenue. Always strive to produce where Marginal revenue=marginal cost. The cool thing about microeconomics is that they aren’t political like macroeconomics are.

    1. johnjcoates

      I agree with everything here, and there should be a follow-up post to further dive into LTV / Churn / Dollar retention. I am curious on other thoughts on Dollar Retention, as explained here, as I am just starting to get my head around it:

  2. awaldstein

    –>Yes I agree with this completely.Understanding this is what business is about.–>Nope I don’t agree with any of it. Simplifying market demand and brand value as measurable by Google Ad words alone is a skewed sample.Feels like the data scientists are redefining marketing as a key word and a click.Curious as the core value of a brand and the issues in challenging the largest ones has never been greater.

    1. fredwilson

      I agree that this is an over simplification and leaves out much that one can do to optimize things. But its a great way to start

      1. awaldstein

        Agree completely Fred.This is something every business should be doing all the time.

  3. Mario Cantin

    MBA Monday is back!

    1. kidmercury

      vintage fredland for sure!

    2. Rob Underwood

      Was just going to say the same! Old school.

    3. fredwilson

      Sort of

  4. kidmercury

    the adwords/analytics optimization cycle works for simple sales cycles, but if you have a complex sales cycle (something expensive, something highly personalized, enterprise sales, etc) or an ad-based business a more involved attribution model is needed. discovering that attribution model may take time which may make unit economics calculations difficult or easily subject to criticism.also, for some products/services even with simple cycles some channels may work a lot better than adwords, and may be subject to lots of external variables (seasonality, earned media….)

    1. awaldstein

      so true.

    2. Jess Bachman

      Agree. Even with “simple sales cycles” its easy to be led astray if you don’t know what your really testing or how to read your results.

    3. fredwilson

      Yes and yes

    4. Girish Mehta

      I took Fred’s post to be relevant to a early stage startup and, as such, a place to start…but honestly in my past experience in a established brand, a lot of my focus was on getting the team to look beyond price elasticity.The fact is that price elasticity isn’t just something waiting there to be discovered; it is intimately related to your brand and, to a degree, you get to shape price elasticity. And Brand Marketing Or Performance/Acquisition Marketing is a false tradeoff.To a degree, you have to be unreasonable about it. Push water uphill.Two things I’d say to my team – “Don’t be a victim of the TAM” and “Don’t be a victim of the Price Elasticity Curve” πŸ™‚

      1. sigmaalgebra

        That curve is somewhere between super simple Econ 101 and Goldilocks fantasy land, e.g., something from the Econ lecture in the Dangerfield Back to School!Ah, my Ph.D. adviser wanted me to take an econ course. So, I got up early, about 6 AM, and made the class on time, 8 AM. I sat in the front, took careful notes, said nothing, did stay awake. The class started with such a curve of quantity as a function of price. But his curve was drawn freehand with no mention of any properties.Convex? Okay, the price of an average of several quantities is less than or equal to the average price of those several quantities.So, as important for doing anything with the curve, e.g., for optimization which somehow I did suspect was coming about the next lecture, after class I asked him what he was assuming about his curve — measurable, integrable, Riemann Integrable, continuous, uniformly continuous, convex, quasi-convex, pseudo-convex, differentiable, continuously differentiable? Later that day I got a message to see my advisor. I did. I was out of the econ course!Such a curve essentially has to exist but after a lot of appropriate assumptions. E.g., what is that curve for transistors? From the early history of transistors to now, any such curve has changed drastically. In particular, early on, the cheapest transistor went for, say, $10. Lower the price to $1, and won’t sell many more but wait a while as customers see more to do with transistors at $1 and will sell a lot more. Now, have the price as some hundreds of million transistors for ballpark $10, but it took a while for the customers to know what to do with hundreds of millions of transistors.So, to make such a curve make sense, about have to specify some short interval of time.And there are more complexities.

        1. Mark Essel

          Enjoyed the comment. While reading I’d magined a 3D shape representing the curve over time, evolving with consumer/product/business analytics/market maturation

          1. sigmaalgebra

            Yup. A 3D surface.Beware economists bearing simplifying assumptions and models (not the arm candy kind — out of their league) — there be monsters, especially when the economists say that the real world should be made to conform to their simplifying assumptions. Good way for them to do more than they did in the Great Depression to start WWII — start WWIII and kill 5+ billion people.Ah, that hapless econ prof didn’t have a chance: Econ had been using mathematical optimization to model how economies work. Several of the Nobel prizes in economics were from this approach, even with just simple linear programming, later with the Kuhn-Tucker conditions (KTC). That’s H. Kuhn and A. Tucker, both long at Princeton (the chair of the committee that approved my dissertation was a Tucker student; the main background math I used in my dissertation was heavily from a star student of E. Cinlar, long at Princeton; ah, once I actually got accepted to Princeton, but only as a math grad student, that is, the easy back door where Dad didn’t need to give $100 million!). E.g., there was such a paper by Arrow, Hurwicz, and Uzawa. Right: Arrow already had his prize; Hurwicz got one later; apparently poor Uzawa has none at all. Don’t worry: Academic econ profs can take ridicule; they are used to it; they get a lot of it; they can take a lot of it; they deserve it, and more.Well, not long before that one fine day in that econ class, I’d written a paper, later published, on some deep, tricky stuff in the KTC that also solved a problem stated but not solved in the Arrow, et al. paper. Ah, on the way I showed that for the reals R, positive integer n, and any closed subset C of R^n, there exists a function f: R^n –> R so that f is 0 on C, strictly positive otherwise, and infinitely differentiable — assume a little less, get a little less than the Whitney extension theorem. Right: For C pick a sample path of Brownian motion, a Cantor set of positive measure, the Mandelbrot set — cute little puppies like those!The KTC have some relatively tricky math: E.g., there is a bunch of stuff from various theorems of the alternative, a bit unpopular in pure math and with some prerequisites also unpopular there. Just to understand the needed constraint qualifications is tricky; understanding the Kuhn-Tucker and Zangwill constraint qualifications is trickier.The chances of that hapless, junior econ prof getting the KTC math straight are slim to none. In front of me, he would be leaving himself open to a Libyan Spanking, i.e., enema, with a bayonet, with a twist. Payback for the Great Depression and WWII and a way to avoid WWIII.If all the economists were laid end to end, it would be a good thing. Their results often fill much needed gaps in the literature and would be illuminating if ignited.E.g., once a Wall Street guy recommended for me Darrell Duffie, Dynamic Asset Pricing Theory. Right, from Princeton. So, I got a copy and started in. The first chapters are really just applied KTC math. Soon I started having some questions! The results were not carefully proved. So, I changed to the other side of the table, started looking for counterexamples. Right: I went through the whole first two chapters and found counterexamples for literally every statement in the book. Then I quit reading the thing. Econ profs stand to have a tough time with the KTC. Not many people know the KTC at all well — it’s tricky and not popular in pure math departments, and nearly no other academic departments are up to the tricky stuff. So, that’s why as a grad student I was able, in two weeks of work, to publish a paper in the KTC and solve something Arrow, et al. missed.BTW — until this thread, I’d never heard of unit economics. But, okay, the definition is simple, trivial. One of my profs warned me — said that I’d need a course in econ if only just to pick up the lingo, their obscure, poorly defined terminology they use to put fog over everything. Ah, a hallmark of low quality work — trivial concepts but with poorly or undefined obscure terminology or three letter acronyms! Easier and more relevant — just be clear on how to make money!

        2. Michael Elling

          Ahhh. And what about network effect on both value and costs where almost all demand elements are infinitely variable. Please professor, help me find the answer! Oh, and is the pricing all you can eat, eat what you want or pay by the morsel? And what is most important, price per unit or revenue per user? Or both?Actually it is quite simple. Optimal ex ante pricing is achieved through multiple iterations of assumptions of marginal supply and demand. But few are willing to, or have the capacity, to attempt this approach.

          1. sigmaalgebra

            Yes, there is a lot of applied math for such a problem, but mostly that math is of the form what you could figure out to do that would be optimal if you had a lot more data than you have any chance of having.So, instead of hoping that math will chew the whole problem in one bite, maybe in some places some math can make a contribution — the math is simple enough and does well enough with reality, the software is simple enough or already done, the needed data is readily available, and the computing is reasonable. There are such cases, but usually have to look to find them.Instead, your empirical iterations are much more promising for the whole problem.But, there, should expect that some experimental design and analysis of variance could give much better results with much less real market experimentation.

  5. TeddyBeingTeddy

    Brilliant post. Thank you Fred. I ace’d Econ 101 (and 102) and still never thought to test/plot price elasticity with my products. This is a gem, almost as valuable as knowing the difference between margin and markup!

  6. BillMcNeely

    Definitely highlighting that the business of business is knowing yours.As interest rates rises and slowly begins to let the air out of free money bubble it will be important to not just have revenue but know to increase decrease it as well as your profit as you pull the levers to affect outcome.

  7. Jess Bachman

    “It is best to keep everything other than price constant as you move through this exercise. For example, don’t change the adwords campaign as you move through this process.”I get the reason for this from an experimental basis, but your ad copy will be priming a certain price. If your ad copy primes a higher price, your charts are going to look different than if you prime a lower price.This type of experimentation is tricky. I’ve seen many people set up experiments wrong, and well… GIGO.

  8. Eliot

    Note: a product doesn’t need to be subscription-based for the customer to have an LTV.

    1. fredwilson

      Can you elaborate?

      1. Eliot

        Depending on ‘subsequent behavior’ of your customers one may or may not need to cover cost of acquisition and COGS in an initial transaction. Cash positive customer acquisition is pretty rare, at least in my (admittedly limited) experience.

        1. JLM

          .You are absolutely correct — nobody knows their real CTA until they actually acquire customers. It also changes over time as you get more efficient at the task.Much of business is taking a risk which has been studied diligently but is still a risk.JLMwww.themusingsofthebigredca…

        2. Eliot

          Bad explanation of what I was trying to get at – if customers repurchase, then you’re worried about LTV. In my little retail business we don’t sell subscription products, but we we work awfully hard to induce re-purchase…

      2. Stuart Willson

        a risk here is that retention rates on non-subscription products (even replenishment) are difficult to gauge, unless one has a large and lengthy data set – and that you acquire customers uneconomically hoping that they will return and justify the original CAC.

        1. Alex Murphy

          And even with a large data set, the retention rate can change based on external factors and turn the business upside down in a hurry.

      3. Alex Murphy

        Two examples:1. Referral oriented go to market strategy. LTV is a function of the contribution from new customers referred by the original. Dropbox as an example.2. Repeat Purchases. Not having any inside knowledge. I imagine Etsy tracks the cost of acquiring a first time customer within the marketplace and then tracks the # of purchases over then next month, year, and so on.

      4. Richard

        Products with high switching costsTeen Apparel

        1. Adam Sher

          NPR interviewed someone senior from LVMH fairly recently and the majority of discussion was around how to maintain profitability and attract design talent in the era of minute fashion. This is where the time from when LVMH introduced new fashion to when H&M et. al copied it was a matter of a week.Based on the speed at which low cost retailers (e.g. H&M, Zara) can introduce new fashion, shouldn’t this switching cost be low?Near the conclusion of the interview, this LVMH executive basically said that he hopes that there will be people who appreciate the craftsmanship and creativity to support (i.e. buy) LVMH et. al. I was surprised by the negative outlook on the industry from an insider.

          1. Richard

            Apparel has done a poor job at coming up with IP strategies

  9. LIAD

    I remember my first economics lesson back in High School.The teacher introduced the subject as the study of a world with “infinite desires and finite resources” – I was smitten from that moment.

  10. LE

    Other things to consider in addition to academic argument as presented above (and this is especially true with the internet).What? “Pigs get fat, hogs get slaughtered” when coming in contact with lack of barrier of entry for your product or service. (Meaning easy for someone to copy what you are offering or selling).In other words just the fact that you can charge $10.00 at an optimum price point might not be what you want to do. Why? Because if you do that others can and will do a napkin calculation and say “oh I can offer that cheaper” and then they will enter the market and drive the price down. So in some cases you want to actually forget about elasticity and simply price the product lower than those optimum points so that it appears less attractive to others because the margin doesn’t appear to be there.

  11. JLM

    .I am stuck in the blizzard in Lubbock and trying to get to the ATX, so I want a lot of sympathy. Not really — big adventure and all.This is a great discussion framer and every entrepreneur/founder should have — master — this and a lot more. This should lead to thinking about how to create the “first demand” — the tryer.Fred’s discussion is tactical — what to do in a force/counter force environment, cause v effect.You have to master that in order to get to the strategic level — what does it take to transform a tryer into a long term, loyal customer.Again, it all starts with that tryer who may be valuable in and of herself but may be even more valuable as she may create “word of mouth” and her value as a tryer, a first mover, may be enormous in the strategic view.While we give lip service to the LTV of a single customer, the bigger company builder consideration is creating loyal customers.JLMwww.themusingsofthebigredca…

    1. Lawrence Brass

      Continuing our monkey talk and considering you might be freezing try some ‘cola de mono’…That would be Monkey’s Tail in english, we have still some in the fridge.Drive safe.

    2. BillMcNeely

      Great point. A lot of entrepreneurs get wrapped up in the consulting/strategy piece and forget they are in a knife fight and don’t learn those skills and get carried off the field on their shield.

  12. Pranay Srinivasan

    What about measuring competiton? Isnt benchmarking also a huge part of this exercise?

    1. Brook Shepard

      Only in as much as it helps you uncover weakness in your own model, kinda? Most similar businesses have similar costs, and the market will bear what the market will bear.

      1. Pranay Srinivasan

        Thats a highly simplistic view especially with Venture Funded businesses who rely on grabbing market share to boost unit economics, and highly competitive spaces can offer a lot of differences between what a market can bear and what a product *should* cost

  13. JimHirshfield

    This doesn’t work for enterprise saas. Thoughts?

    1. Alex Murphy

      Enterprise is centered around ROI. Either because you can radically cut costs or because you can accelerate the business. You have to be able to demonstrate that value with clarity in a way that is easy to understand. Salesforce’s approach toward Total Cost of Ownership was the key to winning the CRM war 10 years ago. There was a real cost to managing CRM software, and there was a soft cost of relying on IT to manage the system. With Salesforce, both went away and there was a cost reduction and the ability to accelerate the business. Win + Win!

  14. Jon Katzur

    Very interesting conversation. Worth noting – in enterprise tech with big sales cost and commensurately high price tags it’s very hard to figure out the price elasticity. It might be different for every single prospect. Regardless, the sample size is so small that for a startup it can be incredibly hard to determine. Still a key point and absolutely something any sales professional and/or entrepreneur needs to think about: just may not be as straightforward as the steps outlined here.

  15. Alex Murphy

    Great thread to start Fred.Some things to consider:- In the equation LTV > CAC + COGs, you have to consider profit and operating overhead as well. A better formula to shoot for at a min is LTV > 2*(CAC+COGS) which gives you an operating margin of at least 50%.- Ad Words is a great place to test price elasticity of “in demand” products and services. However, much of the world (such as the entire Etsy marketplace) is really driven by discovery, not demand- While the unit elasticity curve in theory is a nicely shaped curve from a high price and few units down to a low price and many units, there are some interesting realities depending on your product or service. Most notably that price is an indicator of quality, and for many “things” quality is more important than price at the margin. There are also psychological differences as you test small price chances such as the different results one can get from $48 to $49 to $50.- While testing, there are many variables that can disrupt your control and test groups. Timing, ads clicked on, ad creative, weather, current events, etc. Trying to run a perfectly controlled experiment is extremely hard and generally not possible.In short make sure you make enough money to run your entire business and make sure you keep in mind other factors that may influence the outcomes.

    1. sigmaalgebra

      Really high margins can, at least for a time, cover over lots of little mistakes!

  16. vinit_jain.90

    Fine words of post-holiday wisdom !!

  17. Lawrence Brass

    I wonder if step 5 and 6 could be modified so people not buying could suggest a price they would pay, and then, set the price to the average plus something. This method should converge faster but it may produce fewer points for the curve.

  18. daryn

    What about alienating your best early adopter customers by consistently experimenting with pricing (lowering from what they paid)? Or the snap back once you’ve hit the low end of your pricing and determine what your retail price should be.Glowforge is an interesting study here. They started with a very high “msrp”, but one no one would actually pay due to a huge (50%) presale discount and a very aggressive referral program. Different goal there, but it sure did work wonders with establishing perceived value and providing urgency to get the discount.

    1. LE

      What about alienating your best early adopter customers by consistently experimenting with pricingIs a good point. In other words what do you say to the schmucks? My answer is that “big picture thinkers” typically aren’t concerned with that anymore than Brian Roberts at Comcast cares about what you think about his shitty customer service department or call queue. The reason for that is simple. They are detached and far from the front lines of dissatisfaction and discomfort of customers in a way that let’s them emotionally get beyond that type of feeling (like a bank president who doesn’t have to wait in his own teller line in back of an old lady taking to much time..)

    2. Michael Chou

      I wonder the same question. Particularly for subscription services, existing customers will notice that you lowered the price for the newer ones. I’m curious to learn methods to experiment pricing while maintaining customer loyalty.

      1. aseoconnor

        The smart move would be to drop the price for existing subscribers once you’ve set it. If they notice they’d be pissed, and the good will you would lose isn’t worth the marginal increase in recurring revenue.

        1. Michael Chou

          I appreciate the answer. I should be more clear. Fred’s suggestion is to lower the price to the extreme, and then raise it back to the “sweet pot” (now that it’s discovered). So the process involves both lowering and raising prices. I agree all subscribers’ prices should be equal, but how to maintain all customers’ loyalty with such dynamic pricing?I realize this is probably asking for spoon-feeding. But pricing is a constant question that my company thinks about. Hopefully someone with experiences in subscription pricing could help or point to reference.

          1. aseoconnor

            Hey Michael,If you’re buying traffic and varying the price then I highly doubt someone will know if they’re getting the high end or the low end of the pricing spectrum (especially for an upstart that is relatively unknown).For established brands who want to test the price elasticity of the product I might suggest doing a “white label” and running media off brand to see how consumers react without being worried about tarnishing the brand’s reputation / identity.

          2. Michael Chou

            Thanks for helping!

  19. Salt Shaker

    Price elasticity and LTV (CPI, CPA) are not mutually exclusive. LTV is fundamentally about brand building and building customer loyalty, which can influence PE. Qtrly profit pressure often forces companies to focus less on LTV. Having the necessary capital to legitimately test, discover and measure over time a company’s LTV, which can be dynamic as changes to a brand’s marketing, cost and competitive mix occur, should always be a key and ongoing objective, but it frequently isn’t.

  20. Richard

    There is a Second way to look at price elasticity. One can fix the price and test by changing the cost of good sold (Advertising and marketing)

    1. Campbell Macdonald

      That wouldn’t reveal price elasticity. It is more akin to supply elasticity, though marketing is normally not considered part of the COGS. Maybe there is a “marketing elasticity” or equivalent.

      1. Richard

        I call it conditional price elasticity (conditional on marketing $), you are lowering the price indirectly by raising the marketing $. Remember price elasticity is always conditional. Evian Bottled water at a the US Open vs Evian Bottled water at the market. Evian bottled water as a sponsor at the US Open, Evian bottled water as a non sponsor at the US open.

  21. Dave Root

    How does a company like snapchat or instagram face VC’s and answer their per unit questions if they are, lets say, in their pre-revenue stage?

    1. Brook Shepard

      I would guess that their discussions show or don’t-show intelligence, as they hash through models.

  22. William Mougayar

    I think the part about mentioning the Unit Economics is even more important than figuring out the formulas, because there are variations to these formulas and how you get there; but understanding the basis of the Unit Economics and validating it in the market is important.

    1. sigmaalgebra

      Over time, COGS can go down, e.g., economies of scale, as quantity sold goes up. Or COGS can go up as quantity goes up and causes shortages and costs of inputs to rise, e.g., labor, materials. Lesson: It can be complicated.And, of course, also over time, with progress in technology, COGS can scream down to just the level of the dust on the floor, e.g., cost of sending a bit from the US to Europe between the first undersea cables and now. Or, the Internet started with a lot of T-1 and T-3 lines, T-1 (1.5 Mbps) for a high end Web site and T-3 (IIRC, 45 Mbps) for a backbone link of the Internet.Back in the days of the Internet backbone with T-3 links, the data rate of the entire US long distance voice network was less than 30 Gbps (from some data from one of the Bells, IIRC, one in Ohio), but now can get a little PC adapter card for, what ever at Amazon, less than $1000, for 10Gbps — so one third of the data rate of the whole voice network for $1000-.Lesson: Such curves can change a lot over time.So, for such a curve, one of the assumptions is that we’re limited to just some short interval of time.

      1. William Mougayar

        True. But the basis formulas provided by Fred are a good starting point.

        1. sigmaalgebra

          Yes, fine, but the lesson I was trying to make is that an assumption of such calculations is that we’re talking about only some short interval of time.

          1. pointsnfigures

            measure consistently, consistently measure

          2. sigmaalgebra

            Sure, but what about my suggestion of using Newton iteration when throwing money at Adwords looking for the lowest price?!! :-)!!

          3. William Mougayar

            Got it.

      2. Michael Elling

        It was metcalf (network effect) compounded by moore (processing effect). What few net neutrality IPistas appreciate is that data was a pimple on the voice elephant’s behind in 1990 and it was the government’s mandated interconnection and equal access in the early 1980s that began the 99.99999% unit price decline for all; not settlement free peering. In fact, the latter will be viewed as an aberration that led to more, not less monopoly and stultified growth. With optics and the right sort of interconnection and settlement models and the fun stuff begins.

  23. Brook Shepard

    I’ve expressed sentiments like this before, but as an agency, we go through this exercise with every new client. You’d be shocked – or maybe it sounds like you’d not be shocked – how few understand this.

  24. Salt Shaker

    What companies are truly nailing a freemium biz model? It certainly isn’t the streaming music services, where the conversion rates from free to pay remain relatively low. I think if not done properly a freemium model can undermine both price elasticity and LTV, particularly when there isn’t a significant and meaningful difference between the free/pay offering (which IMO is the case w/ streaming music services).

  25. goldwerger

    There is one more scenario in real life that’s not covered by the regular elstacity curve: when demand increases as you increase price.It’s less common but it’s awesome when you hit it. It’s commonly a result of product positioning, or anchoring of perception. Some players, like Apple, do a brilliant job driving huge demand to products positioned as both essential and premium (so they flip the curve beyond inelastic to outright opposite).Apple is a extraordinary. But I’ve always believed pricing is one of the biggest under-managed levers in most companies. Pricing can have a strategic, not just a tactical, impact on a business.

  26. creative group

    Refreshing to read a contributor expand on what some of the smartest people in the room blog or contribute to this blog. Just when an outsider says oh sh#t eureka on an operational method published someone else follows up and stretches the entire calculation to another business variable.Another reason to subscribe to this blog.

  27. Mohan R

    I like this because it really connects to the fact that companies can set prices but markets are the one that give credibility to the price point.

  28. Brandon Burns

    I love the effort to show a way to calculate this; I wonder if there is a more foolproof way to get untainted numbers.Like you said, you want to keep everything except price a constant. But time and external factors are never constant.As services grow, and get PR, and brand recognition grows — so does conversion rate, since people are more likely to buy something they’ve already heard about. And in the beginning, this change can be dramatic — when I launched Wander&Trade, conversion went from 0.4% to 2.6% over the first six weeks.I think running the tests simultaneously, concurrently instead of consecutively, with different landing pages for different prices, would lead to a better statistical environment, and thus yield more accurate results (not to mention, a faster route to an answer).

  29. Camille Tyan

    In running the experiment you need to make sure that the customers you attract via the AdWords campaign represent your target user. The worst thing that can happen would be understanding the elasticity of a product for the wrong target.

  30. sigmaalgebra

    Ah, the iterative scheme 1-10 could be seen as a bit inefficient, that is, poor use of spending on Google Adwords.But, looking at the curve of quantity as a function of price and the desire to find the lowest price so thatPrice > CAC + COGScould do a binary search. Better still, since the curve is convex and differentiable (just by eyeball), Newton iteration (not nearly new!), e.g., a good way to find square roots, should be much faster. And why faster? Save on spending on Google Adwords!Okay, want to optimize: For that a lot is known.

  31. @MartyCrampton

    Damn! Does that mean my startup needs to make money? So, I guess Mark Zuckerberg, @ev/@jack took-up all the loose (we’ll think abt how to many money later) VC money!? (Does Uber’s “surge-pricing” fit what you’re talking about?)

  32. Patrick Campbell

    There’s an easier way to do this at any stage company for better results and richer data. For context, I’m the CEO of Price Intelligently – our software helps companies measure customer perceived value and price elasticity. Customers include Bigcommerce, New Relic, Wistia, etc.Attempt at establishing ethos aside, let’s walk through the problem with this methodology, and then give a bit of a better one.There are two core issues with the above methodology: 1. you’re missing out on packaging and positioning implications of your pricing, and 2. there are too many lurking variables when it comes to an AdWords/traffic driven test. If you have an extremely basic and straightforward product (an ebook, physical product, etc.), then the above works ok, especially if you have a very searchable product. Anything a bit more amorphous, even a basic software product, will start to suffer from issues where even basic changes in packaging (which features in which tiers) or positioning changes (persona alignment) will potentially cause huge shifts in value. Without going too deep into this, one big thing that we’ve found is that bigger shifts in conversion and lead velocity actually come from adjusting your value metric or shifts in packaging, rather than physical price point.The larger issue with this methodology though is that you open up yourself to lurking variables that can’t be A/B tested easily. Think about a basic SaaS product with 2 differentiable features separated out into 2 tiers and a value metric (# of page views for instance). Testing all of this via an adwords driven test is unrealistic from a statistical integrity perspective, and you’ll likely run into even basic seasonality issues or as @kidmercury mentioned some issues around sales cycle. There’s more to be said here, but the basic idea is that you have 4 main axes that can get complicated quickly – your value metric, your packaging, your price, and your positioning. Testing all four requires a ton of traffic.A better methodology for pricing specifically is to collect some survey data from target customers used the van westendorp methodology. You can also use MaxDiff for some of the packaging and value metric questions. The results will be a much narrower problem scape where you’ll be able to take down your number of tests to 2 or 3, rather than a problematic and potentially endless loop of testing.VW works pretty simply. You ask four ranged questions:1. At what (monthly) price would [product] be too expensive you’d never purchase it?2. At what (monthly)… starting to get expensive, but you’d consider purchasing it?3. At what (monthly)….be a really good deal?4. At what (monthly)…be too cheap that you’d question the quality of it?That data plotted in excel, gives you a visualization that will show you a true elasticity curve where you can measure changes in price’s impact on conversion (sales lost). Here’s an example along with some more information on calculations: http://www.priceintelligent…You can get respondents for your survey from lead lists, AskYourTargetMarket (, or even some other bigger market panelist firms like Fulcrum, Research Now, etc. Sure, this is survey data, but I can get consumer panels for $3/response to a 15 minute survey and B2B panels for $10/response. With some segmentation questions, all of a sudden a few hours worth of work gives you an intense amount of customer research that will then guide 2-4 variations to test. All of a sudden your guess and check problem turns into a much smaller and more actionable scope.There’s more on the MaxDiff side of the house (measuring relative preference for features to guide packaging), but since the post focused strictly on elasticity, I’ll just provide a link to an overview: http://www.priceintelligent…Happy to answer any questions or chat through more pros and cons to the above.

  33. guyAtHockeyBiasDotCom


  34. Roy Bahat

    Love this, with one addendum: many founders should also test *higher* than the highest end of their range. I more often see founders underprice than overprice. Believe in your product! Bloomberg insisted on a high price day one, never discounted.

  35. awaldstein

    Margin on food in channel food goods is honestly pretty much an ice cube in the sun as you know.Pieces are pretty simple except for a few hidden ones (tickets, seasonality of ingredients, mandated channel sales, ever increasing costs of insurance).The hard piece is that it is a volume game and when you don’t own the channel it can break at any time.There is a reason why copacking and outsourcing is growing.

  36. PhilipSugar

    What a great use for a spreadsheet. (I mean that)

  37. awaldstein

    Tough biz and i’m reconsidering my own value add to it honestly.

  38. LE

    yeah but companies that use co-packers aren’t food companies, they’re marketing companies. Hmm. Reminds me of printing brokers vs. people who actually operate the equipment nothing that when it comes time to sell having the iron is actually fairly important to some buyers (rather than just desks in an office..)

  39. awaldstein

    You are a unique case Charlie and it is to your credit and possibly to your advantage.But the net of it is you are either a brand or not and to that end, whether you own manufacturing is not a determinant is all I’m saying.

  40. Rick Levine

    Nice thread for a Monday morning. Thanks for making a place for it to happen, Fred!Lots of gray areas in the conversation about manufacturing. I share some of the biases about brand ownership and control needing to be linked to ownership of manufacturing. (Disclaimer, I’ve been running companies, both food and fashion, that ended up owning manufacturing, usually because there was no other viable option at the time.)Any company that offloads all their manufacturing and sourcing is potentially lowering the barriers to competition. Successful third-party manufacturing requires creating (or at least incubating) process and quality expertise in the outsource vendor(s). Unless you have an IP or other lock on some critical part of the process, your manufacturer can walk away. (There was a sizable venture-funded fashion brand in the sustainable space that had their lunch eaten by a national retailer a few years back. The retailer did an end-around to the manufacturer, and the marketing business died hard. Big player, questionable ethics, nothing personal, just business. :)From the standpoint of both innovation and brand defense, you usually end up with a hybrid execution, moving some manufacturing and sourcing outside, but owning those bits that make you more nimble or more innovative.Tacking back only slightly towards the original post (!) – the catch to customer acquisition, when you’re owning manufacturing and competing against the “marketing” companies with much lower capital costs and less risk, is that you have to find a way to help your potential customers understand the value your in-house manufacturing commitment brings to the product. If customers can’t explain the difference between your product and the out-sourced competition, your business won’t last, and your preoccupation with manufacturing isn’t helping you.

  41. awaldstein

    Nope this is not the same as brokers as in this case the brand creates and markets the product it just doesn’t capitalize the manufacturing.Not the same.

  42. LE

    No I meant the “aren’t food companies” and the fact that Charlie feels apparently a sense of pride that he actually bakes the bread and packs it vs. a co packer.Of course there is the in between state that is bake (or print) some of your products and use others to produce other products which gives you a benefit greater than strictly being either one case or the other.

  43. awaldstein

    Pride in the operation and pride in the product are not the same.The idea that copacking is not a fully managed environment is simply not a reality. Most of the times you do just’s an economic decision determined by how you want to manage your cash.Charlie’s business is uniquely different here. And quite wonderful as a model I might add.Almost no brands manufacture all of their food.

  44. LE

    Pride in the operation and pride in the product are not the same.Disagree. I think that pride in operation and pride in product are the same thing. Not to mention that having an “operation” provides psychic motivation to all parties (employees, owners) that does extend to the quality of the product. (You think there is nothing to walking through a factory? I disagree.)The idea that copacking is not a fully managed environment is simply not a reality. Most of the times you do just that.The above doesn’t make any sense to me.Charlie’s business is uniquely different here. And quite wonderful as a model I might add.Charlie has a few issue to work out (from my personal knowledge) with his production.That said I am not sure why you would say Charlie’s business is “uniquely different”. He bakes and sells bread what is unique about his model exactly?

  45. awaldstein

    You are presuming that people in the business of making stuff, not data, are not in control of their brands if they don’t own manufacturing.Bold statement and elimates the fashion, food and computer Industries as no one owns and operates all of their factories in any of these.Also eliminates franchise businesses as controllable brand extensions.What Charlie is doing is unique.But it has nothing to do with control or pride.

  46. LE

    First I didn’t say that and didn’t assume that.But as far as “owning the factory” I would say if you are a large important brand (like Apple as only one example) and you are buying equipment and all of the production (which they do) for the companies that do manufacturing for you then you are actually “running the factory” and own manufacturing even. Someone else might legally “own” it but you “own” them and control them. This is much different than a copacker who serves 100 masters (and you give them 3% of their business).Other examples where brands own “factories” either dejure or defacto: Automobiles, Cameras, Watches (my assumption you know those high end expensive bullshit swiss watches). Wine (Vineyards my guess). Airplanes and jets. Cigarettes. Coca Cola.Of course car companies do in some cases use other car makers factories.[1] Defacto:

  47. awaldstein

    As you do with copacking.I know this area. I’m right.

  48. LE

    Owning the facility and bank rolling payroll and inventory management has nothing whatsoever to do with control.If you are large. But if you are small it does since your business isn’t (at a small scale) important enough typically to your vendors (depending of course on how much business you give them) to allow you as much control as you need. [1][1] Note that this extends to web resources and services as well. If you run your own servers in colocation, or with T lines, you have way more control than if you are using AWS or someone else’s VPS (Rackspace, Linode, Digital Ocean) because you have to contend with their rules and policies and what they think is important which might differ from what you think. [2] (Not that there isn’t a benefit to doing that in some cases but it does absolutely come with drawbacks which are often ignored…) If you are running servers at Mediatemple and they say “we are retiring our non ssd servers and need you to migrate to other servers” you will do it according to their timing not what is good for you. If you are running your own equipment you control the timing of that so it’s not arbitrary.[2] For example Rackspace has a detailed definition of what types of email they consider spam and can pull your account for sending said email (according to the TOS). If you are running your own equipment you are only subject the rest of the internet thinks is spam which is quite different than what Rackspace might do to protect their network.

  49. awaldstein

    You are incorrect in this instance LE even though your logic is sound.Thresholds drive price more than access.It’s a viciously competitive business and analogs from other industries are not very pertinent.

  50. LE

    Thresholds drive price more than access.You’d have to explain what you mean by that also I don’t see how it relates to what I have said.

  51. awaldstein

    It means that you have access to the model you just pay more if you have less volume as long as you have minimum thresholds.

  52. Richard

    Dealt with a few people in this space, one solution we came up with was supplying restaurants while the retail channels filled the capacity gaps.

  53. Rick Levine

    Heh. “…once you’ve hit the right scale … it becomes a very nice business.” Amen.We’re on the other end of the spectrum. (We make socks.) Almost all our competitors outsource, and their products may be an order of magnitude cheaper to manufacture. We can’t compete on price, ever, so communicating the tangible reasons why we’re different/better is critical.One thing that’s been hammered home for us is that trial drives sales. (Charlie, I suspect you know all about that!) One sale with a little bit of customer care will get us lots of repeat business. If we can get a someone to try on a sock, to touch it, they become a customer. Unfortunately, there are fewer places to sample socks than there are for food, and the process isn’t supported by normal channels the way food sampling is. We’ve taken to carrying single socks, neatly packaged, and letting prospects pick one from an assortment to take home.The bad news is that small-scale, organic customer acquisition is hard, and unpredictable in the short-term. The good news is that the business builds very nicely over time, and a customer from that process becomes an evangelist.

  54. LE

    which are for test marketing new versionsJust be aware that if the “special variety” is a customers first purchase they might not consider the brand on other more stable items if the quality is not good.

  55. Richard

    I don’t think anyone has done a baked goods franchise wholesale business in sometime. This model may make sense to scale bread nationally.

  56. awaldstein

    Really interesting comment.I don’t know bread but in the beverages there is no customer visibility or concern really about manufacturing. Ingredients, brand, local yes but whether you own the liquifieer is irrelevant.Short shelf life products drive an entirely new set of concerns along with it.

  57. Vasudev Ram

    >is that you have to find a way to help your potential customers understand the value your in-house manufacturing commitment brings to the product.What are the issues with doing that?

  58. awaldstein

    How many product do you buy at your grocery store that you know how they are manufactured and who owns the facility?

  59. Vasudev Ram

    – I know it for some – for those products where that info is put on the label. And that is at least some – unless they are blatantly lying. (But that leaves them open to govt. prosecution, so I’m guessing it is not very likely).- I personally try to buy organic and home made / hand made products whenever I can, including verifying those points in person.- I’m not a manufacturer myself, but I personally know some small-scale manufacturers.I asked Rick the question because his phrase:>is that you have to find a way to help your potential customers understand the valuesounded like he thought it was always an issue, not just sometimes.(Edited to change “his:” to “his phrase:”)

  60. awaldstein

    Thanks.Just because you make your product in another person’s kitchen does not mean they are not hand made.

  61. Rick Levine

    Hi Vasudevram!I think I understand your question, but yell if I’m missing it.Offhand, I can’t think of products where the fact of the brand owning manufacturing is important to the consumer. People really don’t care, except possibly as a reinforcement to the brand story.Taking socks as an example, most people have no idea what goes into the manufacturing and distribution of a pair of socks. Yes, it’s probably the most complex garment you wear, and the supply chain is complicated and difficult to navigate, and creating a US manufacturing operation is both expensive and time-consuming.However, the only two things driving a sale, and more importantly a second sale, are that you think the socks look great on you, and that they’re really comfortable.If you’re trying to compare the 12-pack of socks you bought from Walmart for $6 with my $29 pair of socks, and you don’t perceive any tangible difference between them, then I will never get you as a customer. It’s my job to help you come to that perception.The fact of my in-house manufacturing doesn’t track directly to the benefits, unless you’re one of those rare folks who get some satisfaction from knowing that things are made by caring, clever, competent people on this side of the pond. And then, if the socks are ugly and uncomfortable, those tangible factors trump everything else.

  62. Vasudev Ram

    Agreed. But then I’m not sure whether I have misunderstood Rick’s question or your first reply to me.What do you think he meant by:>understand the value your in-house manufacturing commitment brings to the productif it is not what I assumed? I assumed he meant in-house mfg. would imply to the customer, better chances of quality, craftsmanship (whether hand made or mass produced), etc.

  63. Rick Levine

    In our case, the implication of quality from in-house manufacturing isn’t usually enough. Our socks stretch enough to fits men’s larger legs, despite the complex patterns, and that’s because we’ve patented the design analysis techniques to make that possible and rebuilt and reprogrammed the knitting machines to make the capability useful.It’s the resulting comfort, ability to do short runs of many designs very quickly, and our textile pattern design chops that make us competitive. We market the end result, and only peripherally toot our horn about the manufacturing.

  64. Rick Levine

    Completely agree. If I can make chocolate that tastes head-and-shoulders better than any other, or can be completely allergen free, then I have a product.My ownership of manufacturing may get me that advantage, but in very few cases does the fact of ownership get me any marketing slack.Actually, the allergen-free angle is one of those cases where it can be used in marketing. We had people ask very explicitly how we knew we were peanut-free, and the only answer that satisfied them was that we own the plant, run the equipment, and watched their chocolate being packed on our clean machinery.

  65. awaldstein

    Nothing matters unless you make it core to why customers care.Allergen free has nothing to do with whether you own your manufacturing process, it has to do with your process.The net of it in the food business is the relationship between cost of materials, capital float for facility and labor and the price your brand can drive in a system where the retailer takes 30 pts and the distributor somewhat less.

  66. awaldstein


  67. Rick Levine

    (Complete aside – Try calling Dan Hofland at Sungold Foods in Fargo. Very consistent organic sunflower butter, ability to vary milling, and he has the wherewithal to respond to any quality problems.)

  68. Richard

    Are you in the Sock biz?

  69. Vasudev Ram

    HI Rick, Got it now (along with your other explanation below). Makes sense and seems like a good position. Thanks.

  70. Rick Levine

    Yes!> Allergen free has nothing to do with whether you own your manufacturing process, it has to do with your process.Conceptually, very true. Practically, in 2006, it made the business, as there was no other alternative. We either paid to build the manufacturing facility, or we paid someone else’s profit and the cost of the manufacturing facility and the cost of inefficiency of running an R&D operation indirectly.(And if you can find a major retailer that moves organic/natural only taking 30 points, my hat’s off to you. πŸ™‚

  71. Rick Levine

    Yes! Apologies if I didn’t make that clear!

  72. sigmaalgebra

    > access to the modelArnold! I’m shocked, shocked! Arm candy?

  73. Rick Levine

    Yes, AFAIK they grind organic in-house. When I bought from them, we had to be aware of the schedule for changeovers from conventional to organic on their equipment. I think they’re certified by ICS in Medina, ND – look for a cert through their site,

  74. lisa hickey

    Hah. Agree.