S-Curves for Trend Forecasting

Basics

This is an S-curve.

The S-curve is a fun­da­men­tal pat­tern that ex­ists in many sys­tems that have pos­i­tive feed­back loops and con­straints. The curve speeds up due to the pos­i­tive feed­back loop, then slows down due to the con­straints.

When the con­straint is bro­ken, the pos­i­tive feed­back loop ramps back up, un­til it hits an­other con­straint.

Diffu­sion S-Curves

The S-curve pat­tern is quite com­mon in the spread of ideas, prac­tices, and tech­nolo­gies, al­though it rarely looks quite as pretty. The ex­am­ple be­low shows “diffu­sion s-curves”—How a tech­nol­ogy spreads through a pop­u­la­tion (in this case US households

The pos­i­tive feed­back loop in this case is word of mouth, and the con­straints rep­re­sent fun­da­men­tal bar­ri­ers to cer­tain mar­ket seg­ments or growth such as sim­plic­ity, us­abil­ity, scal­a­bil­ity, price, etc.

This cre­ates smaller s-curves around adop­tion among spe­cific mar­ket seg­ments, and larger s-curves that rep­re­sent the over­all mar­ket pen­e­tra­tion of the idea, prac­tice, or tech­nol­ogy.

Evolu­tion S-Curves

In ad­di­tion to Diffu­sion S-curves in tech­nol­ogy, ideas, and prac­tices, there are Evolu­tion S-Curves. Th­ese rep­re­sent the in­crease in the traits of these ideas that make them us­able in more situ­a­tions and de­sir­able for more peo­ple. When you break through a con­straint in one of these prop­er­ties through in­no­va­tion, this can of­ten co­in­cide with “un­lock­ing” a new diffu­sion curve by open­ing up a new mar­ket that wouldn’t pre­vi­ously have used your tech­nol­ogy or idea.

In this case the pos­i­tive feed­back loop is the in­creased un­der­stand­ing and ex­per­tise that comes from diffu­sion of a new in­no­va­tion in your idea or tech­nol­ogy, and the con­straint rep­re­sents fun­da­men­tal as­sump­tions in the idea, prac­tice, or tech­nol­ogy that must be changed through an­other in­no­va­tion to make the idea, prac­tice, or tech­nol­ogy more de­sir­able.

In the ex­am­ple be­low the de­sir­able prop­erty is hard­ware speed. Fun­da­men­tal leaps are made to break through a speed con­straint, and then iter­ated on through the pos­i­tive feed­back loop of in­for­ma­tion and ex­per­tise in­creas­ing from adop­tion. This hits diminish­ing re­turns as the new in­no­va­tion is op­ti­mized, and then a new fun­da­men­tal in­no­va­tion is needed to over­come the next con­straint.

S-Curves vs. Ex­po­nen­tial Growth

Some­times, peo­ple get con­fused and call S-curves ex­po­nen­tial growth. This isn’t nec­es­sar­ily wrong but it can con­fuse their think­ing. They for­get that con­straints ex­ist and think that there will be ex­po­nen­tial growth for­ever. When slow­downs hap­pen, they think that it’s the end of the growth—in­stead of con­sid­er­ing that it may sim­ply be an­other con­straint and the start of an­other S-Curve. Knowl­edge of Over­lap­ping S-Curves can help you model these situ­a­tions in a more so­phis­ti­cated way.

S-Curves Patterns

S-curves be­come quite use­ful when paired with an un­der­stand­ing of evolu­tion­ary pat­terns. They can al­low you to see in a broad sense what’s com­ing next for an idea, prac­tice or tech­nol­ogy. They can pre­vent sur­prises and give you a tool to stay ahead of changes.

There are pat­terns that ex­ist for both diffu­sion and evolu­tion S-curves.

Diffu­sion Patterns

Diffu­sion pat­terns de­scribe com­mon themes that hap­pen as trends diffuse through a pop­u­la­tion. They ap­ply on the micro-level to in­di­vi­d­ual pop­u­la­tion-seg­ments, and on a macro-level to the over­all pop­u­la­tion.

Diffu­sion of Innovation

The diffu­sion of in­no­va­tion de­scribes 5 sep­a­rate stages of a diffu­sion curve: In­no­va­tors, Early Adopters,Early Ma­jor­ity, Late Ma­jor­ity, and Lag­gards. By un­der­stand­ing the traits of each of these groups, you can get a broad idea of what to ex­pect, and how to slow or speed up adop­tion.

The Chasm

The Chasm de­scribes a com­mon con­straint that oc­curs in a mar­ket seg­ment be­tween “early adopters”—who are will­ing to put up with a lot, and “early ma­jor­ity”, who ex­pect a lot. There is of­ten a num­ber of evolu­tion­ary con­straints that must be bro­ken through to bridge this sin­gle diffu­sion con­straint and many new ideas, prac­tices, and tech­nolo­gies get stuck in the chasm for that rea­son.

Evolu­tion Patterns

Evolu­tion pat­terns de­scribe com­mon ways that in­no­va­tions evolve over time to be­come in­creas­ingly de­sir­able. They ap­ply on the micro-level to in­di­vi­d­ual in­no­va­tions within a trend, and on a macro-level to the evolu­tion of trend as a whole.

Wardley Evolution

In­no­va­tions tend to go through four stages—the ini­tial pro­to­type, cus­tom built ver­sions, pro­duc­tized ver­sions that com­pete, than co­modi­tized ver­sions that are all ba­si­cally the same. By un­der­stand­ing where you are, you can un­der­stand the type of com­pe­ti­tion likely to hap­pen, the types of pro­cesses likely to yield im­prove­ments, and large changes that will be needed to stick with the mar­ket.

Sim­plic­ity—Com­plex­ity—Simplicity

In­no­va­tions tend to start out rel­a­tively sim­ple as a new ap­proach to a prob­lem. They be­come in­creas­ingly com­plex to cover more use cases and be more ro­bust, and then be­come sim­ple again as re­fine­ments are made and they’re dis­til­led to their essence.

Dis­rup­tive Innovation

Some­times, in­no­va­tions over­shoot the main­stream pop­u­la­tions needs on a par­tic­u­lar di­men­sion in or­der to be pow­er­ful for a par­tic­u­larly lu­cra­tive part of the pop­u­la­tion. In this case, these in­no­va­tions or of­ten over­taken by sub­se­quent in­no­va­tions that lower the perfor­mance on that di­men­sion in or­der to raise it on other di­men­sions (ex­am­ple: Lower flex­i­bil­ity of a soft­ware product but raise the sim­plic­ity), these in­no­va­tions can then “dis­rupt” the origi­nal in­no­va­tion.

From the per­spec­tive a cur­rent in­no­va­tion, the dis­rup­tive in­no­va­tion ap­pears to start be­low it in the s-curve, but it’s able to gain adop­tion be­cause the par­tic­u­lar perfor­mance fea­ture of that in­no­va­tion is already higher than the mar­ket needs, and the new product com­petes on a differ­ent perfor­mance fea­ture that is not even a tar­get o

(image of how that plays out with S-Curves)

Gart­ner Hype Cycle

The Gart­ner Hype Cy­cle de­scribes a par­tic­u­lar way that the me­dia over-in­flates peo­ple’s ex­pec­ta­tions of new in­no­va­tions in com­par­i­son to how evolved they ac­tu­ally are for a par­tic­u­lar mar­ket seg­ment’s needs.

Win­der­mere Buy­ing Hierarchy

The Win­der­mere Buy­ing Hier­ar­chy de­scribes four differ­ent im­prove­ment fo­cuses that an in­no­va­tion op­ti­mizes over time. First, it’s try­ing to solve for func­tion­al­ity, then re­li­a­bil­ity, then con­ve­nience, and fi­nally price. This loosely maps to the stages of Wardley Evolu­tion.

(image here)

Conclusion

S-curves and s-curve pat­terns are a use­ful tool for quickly an­a­lyz­ing sys­tems, par­tic­u­larly when look­ing at diffu­sion of trends and evolu­tion of in­no­va­tions. They can heuris­ti­cally iden­tify solu­tions and prob­a­bil­ities that would oth­er­wise be quite time con­sum­ing to figure out us­ing some­thing like a full sys­tem or func­tional anal­y­sis.

Hope­fully you find this tool use­ful in your quest to un­der­stand all the things.