Hiring for category experience is an expensive habit in ecommerce

2026-09-14

Category experience is a proxy. It is cheap to verify, easy to explain to a board, and simple to defend if the hire goes wrong, because nobody has ever been criticised for recruiting from a direct competitor.

• Ecommerce
• Hiring
• Recruitment
• Careers

Every so often I come across a job specification for a Head of Ecommerce or an Ecommerce Director which has one sentence that causes more harm than all the other sentences put together. This sentence is generally placed towards the end of the document and is stated as if it's self-evident. It says that candidates must have experience in the jewellery industry or in pet products or in the furniture sector or in fixtures and fittings or in the premium FMCG area. I have been eliminated by that sentence. I have also had to sit on the other side of the table while a shortlist was reduced by half because of it, and have watched as we lost two people who would have been better than the one we eventually employed.

The following argument states that the sentence should appear in most JDs, explains why it was included, and suggests what should replace it. To be completely fair, I also include a section on examples where a cross-sector hire did not work well and what it tells us.

What the filter is really selecting for

Begin by considering the psychology of the individual carrying out the screening, since it is in that area that the sentence originates. When a hiring manager goes through fifty applications, they are carrying out a similarity assessment against a mental image of the ideal candidate. This is referred to as the representativeness heuristic, as Tversky and Kahneman explained: instead of consulting the base rates, people evaluate the probability by looking at how closely a case matches a prototype. The prototype that exists in a hiring manager's mind is based on the individuals who are currently doing well in the company and on the competitors who are generally respected. The category is the most salient feature of this prototype and appears on the first line of a CV, which is the reason it is used as the basis for sorting.

On top of this is defensibility, a less attractive but more significant factor. Research into accountability indicates that when decision-makers expect to have to justify a choice to others, they tend to go for options that are easy to defend rather than those that are most likely to be correct. Hiring is precisely such a situation. When appointing a Head of Ecommerce, the position is visible, costly, and is assessed by people who were not present at the interviews. If the candidate had come from a direct competitor and turns out to be unsuccessful, the decision looks sensible and the failure is blamed on external circumstances. But if the candidate came from a different category and fails, it is the decision itself that is criticised. This imbalance is clear to all present.

The result is that the pool shrinks to include only those individuals who are already following the standard playbook in your industry. You end up paying for the typical practice within your own field, at a higher price since all the other companies in that field are competing for the same twelve people. The commercial implications appear in two areas. The first is purely mechanical: it takes longer to fill the position, there is a longer period with the post vacant, and the final salary is higher. The second one is something that no one ever measures, and that is the lack of any practice arbitrage; all the improvements I have seen that actually increased revenue per visitor came from a sector that had previously solved the problem more effectively.

What the research says, and what it does not say

The most useful single piece of evidence is a meta-analysis carried out by Chad Van Iddekinge, John Arnold, Rachel Frieder and Philip Roth and published in Personnel Psychology in 2019. This study combined 81 separate samples which had looked at pre-hire experience and its relationship to performance and turnover in the new organisation. The corrected correlation with job performance was 0.06, with training performance it was 0.11 and with turnover it was 0.00. Simply put, the amount of relevant prior experience that someone has tells you almost nothing about how they are likely to perform when they join.

The paper in question is seven years old and that is beyond the point at which I am prepared to quote it without verifying it ourselves, so I looked for any research that had challenged it. I couldn't find any. The finding has been widely cited, and the subsequent literature has tended to explain the result rather than to dispute it, usually by saying that experience measured in years and job titles is a crude stand-in for what is actually important, namely what the person has learned and how deliberately they have done so. This represents a refinement of the original argument and happens to strengthen the case here: the issue with the category filter is that it measures the crudest form of experience that is available.

Now let us consider the counterweight since there is one and it warrants a proper discussion. Boris Groysberg had spent many years researching what occurs when top equity analysts transfer from one bank to another, a study which was published in 2010 under the title Chasing Stars and also appeared in a paper co-authored with Lee and Nanda in Management Science in 2008. The main conclusion from this is that the stars deteriorate in performance after they move, and this decline continues for several years. Such a situation would not arise if performance were completely transferable.

Looking at the details, the finding then supports the position I am arguing for instead of opposing it. It is the firm that fails to keep pace with the analyst. Specifically, the research infrastructure, the colleagues, the internal relationships, and the reputation which had enabled access to various parts of that institution. Groysberg found that analysts working in a team could maintain their level of performance and that those who moved to companies with stronger capabilities fared better than those who moved to companies with weaker ones. The economic rationale for this is old and well established, dating back to Gary Becker's distinction between general human capital, which belongs to the individual, and firm-specific human capital, which remains with the firm. Category knowledge is squarely general human capital. It is the one thing in the whole bundle that genuinely does travel, because it is stored in the person rather than in the organisation, and it is the thing the job spec is filtering on.

The four things that actually differ between categories

When I assume control of an ecommerce business in a category in which I have no previous experience, I have to go through a substantial period of learning. I need to be frank about the nature of this learning curve, since attempting to pretend that it doesn't exist would make the rest of my argument appear naive. From my experience, the curve can be reduced to four factors, all of which are contained in the business's own data.

Basket frequency comes first; it's about how often the same customer returns and whether the interval is measured in days, months or years. This factor has a major influence on all the subsequent decisions. Such as whether email acts as a revenue channel or as a nuisance, whether a loyalty programme can cover its costs, what an appropriate payback period for acquisition should be, and whether the homepage is designed for returning customers or for people who are new. Contact lenses and engagement rings are at opposite ends of this spectrum, and no other aspect of them is as important.

Replenishment cycle is the second, and it is distinct from frequency. A consumable has a predictable burn rate, which makes subscription viable and makes a late delivery a stockout in someone's bathroom cabinet. A durable has no burn rate, so the reorder trigger has to be constructed out of life events or seasonality. Get this wrong in either direction, and your CRM programme fires at the wrong moment.

The return rate ranks third. Fashion operates at a level that would cause a furniture business to fail, while furniture operates at a level that would be a disaster in the grocery sector. The rate itself isn't very interesting because of what it has on the unit economics and the customer's mental model; in a category with a high return rate, the customer regards the order as provisional and buys three in order to keep one, so part of your conversion rate is fictitious and the contribution per order is the only figure worth looking at.

The fifth factor is the duration of the consideration period. It is the length of time that elapses between the first point of contact and the actual purchase which determines whether your role is to close the deal or to remain available. If the consideration period is short, seamless checkout becomes the most important consideration; but when it is longer, customers are given good content, comparison tools, saved baskets, and patience, and anyone who tries to force a sale is punished.

There are four variables; they can be read out of the data within two weeks by anyone who knows what to look for. Compare this with the list of things that are the same in every category I have dealt with. Such as what a shopper does when the search box gives no results, what occurs in their mind when the price at the payment page is higher than the price in the basket, the effect of a missing delivery date on purchase intent, whether the returns portal defaults to an exchange or to a refund, and how stock scarcity messaging behaves when it turns out to be accurate. That list represents the machinery of ecommerce, and it is based on general human psychology, since it doesn't matter what is in the box.

Five people, and what each one shows

When Alexander Lacik began working for Pandora in April 2019, he had no experience in the jewellery sector; prior to joining Pandora, he had been employed by Britax, a company that produces child car seats, and had previously managed Reckitt Benckiser's North American division, a business worth more than 3.5 billion dollars, having spent thirteen years at Procter & Gamble. As Pandora itself stated, Lacik was a consumer marketer and brand architect and had been given the task of carrying out a turnaround programme which was already in progress. In 2025, the company earned a revenue of DKK 32.5 billion. The ability that had been recruited related to brand engineering in a competitive consumer market, and this ability has nothing to do with jewellery.

Dave Lewis joined Tesco in 2014 from Unilever, making him the first chief executive in the company's history to have come from outside Tesco, and he took up the post at a time when the business was in its worst crisis. The kind of knowledge he had, namely that coming from the supplier side, proved to be precisely the viewpoint that was lacking in a grocery business whose commercial arrangements with its suppliers had gone wrong. A person with twenty years' experience in the UK grocery retail sector would have regarded those arrangements as normal, since within that sector they were in fact normal.

Seb James ran Dixons Carphone and then went to Boots UK in 2018. Electricals to pharmacy and beauty is a long way in category terms. In operating terms it is the same problem: a large estate of physical stores with declining footfall, a digital proposition that needs to become the growth engine, and a range that has to be rationalised without damaging the reason people walk in.

The John Lewis Partnership provides the cleanest natural experiment I know of, because it ran the test twice. Sharon White came in as chairman in 2020 from Ofcom, having previously been a senior civil servant. Jason Tarry took over in 2024 from Tesco, where he had run the UK and Ireland business. Both were outsiders to the John Lewis Partnership. Only one of them was an outsider to retail operations. The turnaround White was hired to deliver was still unfinished when she left, and I would not attribute that entirely to background, because the pandemic and the collapse in department store footfall would have tested anybody. What the two examples show is the fact that the category filter eliminates the distinction between functional experience and category experience, the two being different things, and the sentence in the job specification generally excludes the wrong kind.

Which brings me to the failure everybody in retail quotes. Ron Johnson built the Apple Store and went to JC Penney as chief executive in late 2011. He removed the coupons and the constant promotional cycle and replaced them with everyday low pricing, on the reasoning that customers would prefer honest prices. Sales fell by roughly a quarter in his first full year, and he was gone by April 2013.

What the Johnson case actually proves

It is tempting to file Ron Johnson under proof that you should hire from within the category, and that reading is wrong in an instructive way. Johnson did not fail because he lacked knowledge of mid-market American department stores. He failed because he changed the pricing mechanism that his existing customer base had been trained on for decades, all at once, with no test.

The mechanism which he came across is called the reference price, this being the mental standard that a shopper uses when deciding what a thing should cost, the standard being formed on the basis of the amounts they have previously paid and of the prices which they see next to the item in question. In the case of the JC Penney customer, their reference price was the discounted price obtained via the use of a coupon, and the coupon itself gave the impression of having won a small prize. When the coupon was removed, the perceived price of all the items in the store rose even though none of the price tags was altered, and at the same time the minor reward that had made the visit worthwhile was taken away. That situation is a classic example of a failure, and any experienced professional involved in pricing would have picked it up, no matter what category they were responsible for.

The main point here is that it's about transferring a particular practice rather than transferring the person. Johnson took a pricing model from a brand that had complete control over its supply, no record of giving discounts, and was extremely desirable, and then applied it to a brand that had none of these characteristics. The ability which Johnson does not possess is referred to as analogical transfer by cognitive psychologists, a field which Gick and Holyoak have researched since 1980; this involves the capacity to tell which features of a solution are essential and which are only superficial. Individuals who are good at switching from one product category to another possess this kind of ability. They adopt the mechanism but discard the packaging. It is this capability, rather than experience gained within a category, that one should seek.

What the outsider actually brings you

What turns this from a fairness argument into a commercial one is this: different categories deal with the same problems at different speeds and in a different order; a person who has worked in one of the fast-moving categories therefore has a ready-made solution to a problem which your category is still discussing.

The case of subscription and replenishment processes is clear-cut. With regard to pet food, contact lenses and nappies, it was necessary to create methods for predicting when to reorder, to introduce pause and skip options, and to take into account the financial effects of a subscriber who orders four days in advance, since in these cases a subscription which miscalculates its cycle results in losing the customer entirely. Beauty and supplements are now doing the same thing, mostly starting from scratch and mostly getting it wrong. A person who has come from the pet food sector knows that having a pause function improves retention rather than causing a loss of revenue, and can show you the cohort data that proves this point; it is the repeat purchase rate and revenue per active customer that are affected.

Size, fit and the economics of returns are another. Fashion has spent fifteen years and a great deal of money learning how to reduce returns without reducing conversion, through fit guidance, review-based sizing and the deliberate decision to accept a higher return rate on some lines and not others. Furniture and homeware are at the start of that curve with a far higher cost per return. A fashion-trained ecommerce lead moving into furniture brings a working model of return rate as a managed variable. What moves is contribution per order.

Long lead time communication runs the other way. Made-to-order furniture businesses have had to become good at telling somebody that their sofa is eleven weeks away and keeping them content for those eleven weeks, because the alternative is a cancellation and a warehouse full of a fabric nobody else chose. Bespoke jewellery, bicycles, kitchens and anything with a configurator have the same problem and generally handle it worse. What moves is the cancellation rate on the highest value orders in the business.

Product page density is due to the electrical aspect of the industry. In the case of consumer electronics, companies had to provide specification comparisons that were easy for people to read since customers actually carry out cross-referencing, and it was less expensive to deal with the question on the page than over the contact centre. Nowadays, the premium FMCG and beauty sectors are in the same situation as a result of increasing ingredient literacy, with product pages being designed around mood rather than to answer questions. What matters is the contact rate per hundred orders and the return rate for items not as described.

Every one of those is a mechanism with a metric attached, and every one of them arrives through a person who worked somewhere else. You cannot buy that from a candidate who has spent their whole career inside your category, for the straightforward reason that they have only ever seen your category's answer.

What to put in the job spec instead

There is actually a gap when the sentence is removed, since the concern in question is valid. The hiring manager is by no means stupid; they are concerned that a person might turn up, not understand the customer, and then spend nine months learning at the company's expense. The appropriate course of action is to test for the very thing they are worried about.

I would put the four variables straight into the interview. Give the candidate your return rate, your repeat purchase curve and your average time from first session to order, and ask them what those three numbers together imply about where the acquisition budget should sit and what the email programme should be doing. A person who can read those numbers will tell you things about your business inside twenty minutes. A person who has been in your category for a decade and cannot read them will talk about the products.

To test the candidate's transfer ability directly, since that is the area in which Johnson fell short, ask them to explain something that succeeded in their most recent business and then to state what would have to be true of your business in order for that thing to work here. You should listen to see whether they are able to distinguish between the structural and the superficial features without having to be told to do so. This is, in my opinion, the single most useful question there is for assessing someone being hired across sectors, and it is just as effective when the hire is within the same sector, where the response usually shows that the person has only ever copied.

Where a genuine category requirement does exist, keep it. The only honest exception is where regulation is involved. In areas such as pharmacy, alcohol, financial services, age-restricted products and any other field that requires a licence, there is real compliance knowledge which takes a long time to obtain and where mistakes have serious, irrecoverable consequences. If you state that requirement precisely in terms of knowledge of the particular regime, it will be of useful value; but if you describe it in terms of general experience in our sector, it will once again carry out the same lazy job as before.

The part that is uncomfortable

There is a version of this argument that is self-serving when it comes from someone who has been screened out, and I would rather name that than pretend it is not there. So here is the test I apply to myself. If category experience were worth 0.06 and something else were worth considerably more, what would that something else be, and can I demonstrate it on demand?

As far as I can judge from the same research into selection methods, structured work sample tasks and cognitive ability forecast performance many times better than prior experience does. This is a standard that the candidate has to meet, not an excuse. It means having the ability to look at a company's figures without any prior knowledge and then say something useful about them, which is a more demanding interview than the one in which you describe what you did at a competitor. It also implies that the hiring side has to put in more effort, since a structured task requires preparation and a CV scan only takes four seconds.

That is the trade. The category line in the job spec is a device for making a difficult judgement cheap and defensible. It works, in the sense that it produces a shortlist quickly and nobody gets blamed. It also guarantees that you will hire your sector's average practice and never find out what the sector next door solved three years ago.