Helmer's 7 Powers

Hamilton Helmer, a Stanford-trained economist who spent four decades advising companies (Netflix among them) and running a strategy-based fund, wrote 7 Powers (2016) to answer a question Greenwald leaves half-open. Greenwald tells you whether an industry has walls. Helmer interrogates each firm-level claim of advantage, one at a time, and the seven advantages that follow are his. So is the two-part gate every claim has to clear, which he calls Power: the conditions that let one firm earn more than its rivals, and keep earning it. Both halves have to be present.

What are the benefit and barrier tests?

The first half is a benefit. Something in the cash flow has to improve, through one of three channels: the firm charges more, spends less per unit, or invests less to produce the same output. If you can't point at the line item that moves, there's no benefit. There's a slide.

Then comes the second half, a barrier, and it's the half that kills most claims. A benefit that a competent, well-funded rival can reproduce gets competed away. So don't ask "is this good?" Ask "why has nobody copied it?" Greenwald's asymmetry test asked that of an industry. Here you ask it of one firm, one claim at a time. Helmer sorts the answers into two kinds. Sometimes a rival cannot copy the advantage. Imitation is impossible, or so expensive that attempting it destroys the attacker. Sometimes a rival perfectly well could copy it and rationally declines, because copying would cost more than conceding.

Helmer's shorthand for the gate is "me too won't do," and behind that phrase is his deeper claim that power begins with invention. Something has to be new (a product, a process, a business model). An offering assembled from what everyone already does gives customers no reason to pay more. It also gives the firm no cost advantage its rivals don't share.

Use the gate destructively. It's no accident that the app's power-mapper drill makes "not a power" the most common correct answer. Benefits without barriers are everywhere: great culture, better service, faster delivery, this year's superior product. Every one of those is operational excellence, which Greenwald already sentenced, and it gets arbitraged at the speed of imitation. So write two columns in your dossier for every claim. In the left column, say what it does to cash flow. In the right, say why a competent, funded rival won't have done the same within three years. Most claims die in the right column.

How are the seven powers grouped?

How we teach the seven, and why not in Helmer's order. 7 Powers numbers them one through seven. We group them instead by the barrier that does the work. It's the barrier that decides a claim, and grouping this way turns seven things to remember into four. Read the barrier first and the seven organise themselves. Two of them make share too expensive to buy, and two make the counter-attack destroy the attacker. Two more are made of elapsed time that can't be purchased, and the last one rests on somebody else's permission. Helmer's names for the seven are worth keeping, because they're how the profession refers to these ideas and you'll meet them in every strategy conversation. The definitions, the tests, the arithmetic and the Indian cases below are ours.

For each: mechanism, benefit, barrier, the canonical case, an Indian echo, and the false positive that most often wears its costume.

Family 1 · the barrier is that buying share costs more than the share is worth

Scale economies. The mechanism: serving one more customer costs less when you already serve many, because a cost that doesn't grow with volume is being spread across a wider base. Content budgets, R&D, a distribution network and advertising all behave this way. Purchasing leverage and delivery density, Greenwald's local geometry again, add to the effect. Its benefit is a lower unit cost than any smaller rival can reach, which the leader can bank as margin at the same price or hand to customers as a lower price it can still afford. Its barrier is volume. To reach the leader's costs the challenger must first reach the leader's volume, and it can only buy that volume by running losses for years. Meanwhile the leader, whose cost advantage doesn't expire, can reprice at will to make the campaign negative-NPV and simply outlast the attempt. A rational challenger doesn't start. Helmer calls the stand-off that results the leader's "surplus."

Helmer's canonical case is Netflix originals. A $100m show, spread across ~280m subscribers, costs about $0.36 per subscriber, while a rival with 30m subscribers pays $3.33 for the identical asset. So Netflix can outbid everyone for the show and still run cheaper per head than the loser of the auction (illustrative arithmetic; the mechanism is the point). Our Indian echo is Asian Paints' distribution, the same power we already met in Greenwald's vocabulary. Watch for the false positive, which is bulk in a business with no fixed costs to spread, such as a large trading company or a big EPC firm. Nothing is being amortised, so nothing improves. Watch also for size in the wrong geography. Volume that isn't on the contested fixed cost is spectator volume.

Network economies. The mechanism: each user who joins makes the product more useful to every user already there, so the same service is worth more inside a bigger crowd. It's the demand-side twin of scale economies. Unit cost may not move at all. What rises is willingness to pay. Its benefit is that the leader can charge more, or monetise more, than a rival whose product is feature-for-feature identical, because the network is the product and the leader's is objectively better for the next person to arrive. Its barrier is the cost of buying share again, and it's steeper this time. A challenger has to compensate people for joining an emptier network, through subsidies or free service. Once the market has tipped, no subsidy is large enough, because you'd be paying people to walk away from everyone they came to find. What follows is winner-take-most, but only within the network's boundary, and the first moat module taught you to find where that boundary is: geography for ride-hailing, category for marketplaces, the social graph itself for social.

LinkedIn is Helmer's canonical case. BranchOut raised venture money, built a comparable product, and was dead inside two years. Its product was matchable. LinkedIn's installed base wasn't. In India, look at Naukri (Info Edge). It has held two decades of recruiter-and-candidate liquidity against repeated well-funded assaults, global majors among them. Those assaults dented the position without displacing it. Its share of white-collar listings and its operating margins (illustrative: 50%+ in recruitment, check the segment disclosures) are the benefit made visible. Beware the false positive "we have a community." User counts aren't a network unless you can name the mechanism by which user N+1 is better off because of users 1 through N. Most products that use network language have an audience, and the difference matters. An audience is rented. A network is owned.

Family 2 · the barrier is that copying would wreck the copier

Counter-positioning. The mechanism: a challenger arrives running a better business model, and what that model destroys is the incumbent's most profitable line. So the incumbent watches, understands perfectly well what's happening, and declines to follow. This is Helmer's signature contribution and the purest form of the "would not" barrier. Its benefit is that the new model is simply better, at lower cost, or fitted more closely to what customers want, or both. Its barrier is the incumbent's own profit and loss statement. Matching the challenger means cannibalising the revenue line that pays for everything, so the incumbent's choice to milk the old model and concede the new one is rational and correctly computed. Stupidity and blindness have nothing to do with it, and that's how it keeps happening to well-run companies. Helmer's canonical case is Vanguard, with index funds at a few basis points against Fidelity's active funds at ~1%+. For decades Fidelity's best move in any given year was to keep collecting the active fee and lose a little more of the future, because matching Vanguard meant vaporising its own economics that same quarter.

Worked example: why an incumbent won't match a challenger

The incumbent's dilemma, computed. Here's the arithmetic that turns "won't" from psychology into NPV, run on a synthetic Indian broker. It's modelled on what flat-fee discount broking did to full-service houses after ~2015 (Zerodha's ₹20-per-order model against percentage brokerage). Heritage Securities (synthetic) has 10 lakh active clients, percentage-based brokerage revenue of ₹1,200 crore/yr and EBITDA of ₹480 crore. A flat-fee challenger appears. Matching its pricing would cut Heritage's revenue on the same activity to ~₹300 crore, against a branch-and-relationship-manager cost base of ~₹720 crore that can only be dismantled over years. Model both options over five years at the 12% hurdle:

Check the discounting yourself: −200/1.12 − 80/1.2544 + 40/1.4049 + 80/1.5735 + 100/1.7623 ≈ −106, and the harvest line discounts to ≈ 1,330. The board chooses B, and B is correct on these numbers, this year, next year, and the year after. Now add the agency layer (the CEO's horizon is four years, and the collapse lands on a successor). Add some cognitive dressing too ("those customers aren't our segment"), and you get Helmer's observed sequence. The incumbent ignores, then ridicules, then panics, then capitulates. By capitulation the challenger has built scale economies and a brand of its own, so matching no longer works even if it's executed perfectly. That's the Indian broking record. Full-service houses launched their own discount arms years into Zerodha's rise, by which time the active-client league table had already inverted (pull the NSE active-client series and date the capitulations yourself). Blockbuster against Netflix's no-late-fees subscription is the same play. Late fees were reportedly a material slice of Blockbuster's profit, and that's the line the new model deleted.

Two cautions for you as the analyst. (1) Counter-positioning is a challenger's power, so a mature leader claiming it is telling you a story. Helmer's timing test, coming up shortly, catches this. (2) It's temporary by construction and expires at capitulation, so the thesis has to name the next power the challenger is building with the time it bought. For Netflix that was scale economies, and for Vanguard it was "scale economies shared." For Zerodha it's scale plus a cost culture that must now survive regulation and maturity. A cheaper price with the same model is the false positive. That's a price war you started, and it has nothing to do with counter-positioning.

How do switching costs work as a power?

Switching costs. The mechanism: a customer who leaves you pays a price for leaving, in money, in effort, or in risk, and that price is a discount your rival has to fund before they can even draw level. The first moat module gave you the taxonomy of what that price is made of (financial, procedural, relational). What belongs here is the two-sided arithmetic. Its benefit is pricing power over the customers you already have, on everything they buy next. Renewals, modules, consumables and upgrades all price above the open-market level, because a rival has to offer a better product minus the customer's full cost of moving. Its barrier is that poaching requires the attacker to pay that cost on the customer's behalf, through discounts, free migration, or funding a parallel run, and the bill usually destroys the economics of the attack. When both sides are equally locked, both sides stop attacking. Some markets look strangely peaceful for that reason.

SAP is the canonical case. Ripping out an ERP means re-plumbing the company's nervous system, and the risk of the transition failing kills most business cases before price is even discussed. India's echo is core banking software (TCS BaNCS, Infosys Finacle), plus something more homely, Asian Paints' tinting machine on the dealer's counter. That machine is a physical object, financed, installed, and trained upon, and it turns a shopkeeper's ordinary inertia into a cost you can count. Two warnings. First, the benefit exists only where follow-on sales exist. Locked-in customers with nothing left to buy are a museum, not a moat. Second, switching costs say nothing at all about new customers. Those customers are level ground for the locked incumbent and the challenger alike, so watch share-of-new-wins as the erosion tell. One false positive is high retention in a category where inertia is industry-wide. Retention is the symptom. Find the actual cost a leaver would pay, or the first credible rival who ships a migration tool takes the base.

The advantages made of time, and the one held by permission

Family 3 · the barrier is that time cannot be bought

These two are why a well-capitalised entrant holding a superb plan can still be a generation away from the position it wants. Money buys most inputs. It doesn't buy elapsed years.

Branding. The mechanism: buyers pay more for a good they know to be identical, because of what they already believe about who is selling it. This is Helmer's partial rehabilitation of the word Greenwald demoted, the disagreement with Greenwald that we flagged as one to hold with both hands. Two mechanisms produce the effect and no others. One is feeling. The good simply feels better, so the Tiffany box makes the identical stone worth more (Helmer cites a striking price gap for like-for-like diamonds), and the same machinery runs through gifting, status and identity categories. The other is reassurance. A known name removes a risk the buyer has no way to price, which is what Johnson's does on an infant and what a Tata bond does in the middle of an NBFC panic. Its benefit is a price premium, and nothing else. If the famous name can't charge more, or can't win at the same price while spending less to sell, there's no branding power here, whatever the awareness surveys report. Its barrier is that the brand is the residue of a long and consistent history, and there's no express lane through time. A challenger has to spend years signalling flawlessly and narrowly to build the same belief, and dilution waits at every line extension and every discount along the way. Notice the self-restraint buried in that barrier. Brands die at their owners' hands, through extensions and promotions, far more often than they die under attack.

Tiffany is the canonical case, and Tanishq is the Indian echo. Jewellery is a category whose economics turn entirely on uncertainty about purity and buyback, and Titan attacked that uncertainty with the Karatmeter and two decades of assay-backed consistency. What came out is a visible premium over the family jeweller for the same gold (illustrative; check Titan's jewellery-segment margins against regional listed peers). There are two false positives. One is fame with no premium attached. That's Greenwald's Mercedes point, and it covers most airlines and most consumer electronics. The other is B2B "brands," which are usually reputation resting on switching costs. That's a perfectly real advantage. Just grade it as switching costs and not as branding.

Process power. The mechanism: a company acquires a way of working that produces lower cost or better quality, built out of thousands of small corrections nobody wrote down, and no rival can copy it because nobody can specify it. It's the rarest of the seven and the only one that exists entirely inside the walls. Its benefit is a cost or quality advantage that comes out of accumulated, interlocking routine no organisation chart captures. Its barrier is time again, this time with opacity on top. The process was evolved and never designed, so it can't be written down, bought, or copied by watching. It can only be re-evolved, over a comparable stretch of years. Toyota is the canonical case. In the NUMMI joint venture Toyota opened everything to GM, plant floor included, and GM still needed the better part of two decades to approach the performance of the Toyota production system. Full transparency didn't close the gap. That's the barrier proving itself in public.

India's first echo is Divi's Laboratories' process chemistry. Customers audit its decades of yield-and-cost optimisation in complex APIs at close range, and competitors still can't match it on cost (illustrative; read the gross-margin gap against generic-API peers). Then there's Asian Paints again. Its forecasting-and-replenishment engine started with a mainframe bought in the 1970s to forecast paint demand SKU by SKU, and the company has been compounding that data process for the fifty years since (check the company histories; the moral survives rounding). As for the false positive, it's "operational excellence." This year's efficiency drive is a benefit with no accumulated time behind it. Test for duration. If a determined rival holding full information needs less than ~a decade to match what you do, call it operations and expect it to be matched.

Family 4 · the barrier is that someone else holds the key

Cornered resource. The mechanism: the firm holds something valuable that rivals can't obtain at any price the economics of their business would allow. A patent. A mine sitting at the bottom of the cost curve. A licence. Occasionally a singular team, and that's why Helmer's canonical case is Pixar's Brain Trust, a cornered resource that walked in wearing shoes. Its benefit is whatever the resource itself yields, whether that's a protected molecule's pricing, the cheapest ore in the market, or the only permission to operate. Its barrier is somebody else's say-so. Property law, a regulator, or personal loyalty stands between the rival and the asset, and no cheque clears that obstacle.

It's the most casually claimed of the seven, so put every candidate through five questions before you allow it.

• Is the resource specific, this asset and not capital anyone could deploy?
• Was it acquired below what it's worth? This question does most of the work. A licence won at full price in a competitive auction hands the rent to the seller and corners nothing. Indian telecom spectrum is the national case study, and it's why "we own spectrum" was never a moat.
• Does the value travel with the resource? If it only works inside one company's routine, you're looking at process power instead.
• Is it still producing a benefit today?
• Is it large enough on its own to explain the return gap you're trying to account for?

India offers two echoes. Coal India's legacy allocated blocks were granted and never auctioned. That's what lets them clear the below-value question. But the authority that granted them can reprice them, so Greenwald's leased warning applies in full. Pharma Para-IV exclusivities are 180 days of cornered time, and Phase 5 builds on them. Its false positive is the line "our people are our moat." People are a cornered resource only while their loyalty holds and only where the value demonstrably travels with them. Attrition takes this one to zero at the speed of a resignation letter.

The seven powers in one table

The seven on one card (the exam deck harvests this table):

Read the barrier column down the page and the whole taxonomy collapses to four ideas, which is the point of arranging it this way. Share can be too expensive to buy. The attack can destroy the attacker's own economics. Time can be unpurchasable. Or the thing can simply belong to somebody who won't sell. That same compression explains why the four lenses converge as often as they do. These barriers are Greenwald's asymmetries, itemised at the level of a single firm.