Why I do not believe in consulting for growth and technology

Advice and the work are two different products

A consultant comes in, analyses your business, writes up what should change, and leaves. You keep the document and the invoice. Turning the document into results is your job.

That is the deal I do not believe in for growth and technology. I do not consult. I implement and optimise. Strategy matters, and I am not against it. But strategy without execution is just a document, and execution without accountability is just busy work.

This article is my case for that view, made as fairly as I can make it, with invented businesses and figures so I can show every step. I show where advice on its own is worth every cent, why it so often fails to turn into growth, what the research says about the gap between knowing and doing, and how to put numbers on the choice between paying for advice and paying for the work.

Worked example: meet the shop

Sorcha runs Spanish Arch Tea, an online tea shop in Galway at spanisharchtea.example. She sells loose leaf tea, gift tins and a monthly subscription. Her shop turns over about €30,000 a month, and she keeps 40 cent of every euro after the tea, the packaging, delivery and payment fees. That 40% is her contribution margin.

Two things bother her. A good few people start the checkout and never finish it, and she does not trust her analytics, because Google Ads and her shop report different numbers for the same month. She has a part time developer, Tadhg, who works for her two days a week, mostly keeping the shop running.

She has two offers on the table. One is a consulting engagement: a specialist reviews the checkout and the tracking and delivers a report with recommendations. The other is a build engagement: someone reviews the same things and then makes the changes themselves.

What you are actually buying in each deal

The two offers can look alike on paper. Both start with someone looking hard at your business. The difference is where the work stops, and who is left holding it.

In a consulting engagement you buy a diagnosis and a set of recommendations. The deliverable is knowledge: what is wrong, what to do about it, in what order. The consultant is paid for the quality of that knowledge, usually by the day or by the project, and gets paid whether or not anything changes afterwards.

In a build engagement you buy the change itself. The deliverable is a working checkout, clean tracking, a campaign that runs. The person doing it has to get past every obstacle the report would have skipped over, because the job is not done until it works.

That difference decides who carries the risk that the plan never happens. Under consulting, you do. Under a build, the person doing the work carries most of it, because a change that does not work is visibly unfinished.

An example: the consulting offer to Sorcha costs an invented €6,000 for four weeks and ends with a 40 page report. It recommends removing two checkout steps, adding a guest checkout, fixing the consent setup so that Google Ads receives the conversions it is allowed to receive, and lining up the shop's order numbers with the ad platforms. The build offer costs an invented €9,000 for the same scope and ends with those four changes live on spanisharchtea.example. The first leaves Sorcha with a to do list for Tadhg. The second leaves her with a different shop.

When advice on its own is worth paying for

I would be overstating my case if I said advice is never worth buying. It is, in at least three situations, and they have something in common.

A second opinion. You are about to make a big decision that is expensive to reverse, a new platform, a new market, a large hire, and you want someone with no stake in the answer to check your reasoning. The work afterwards may be done by someone else entirely. What you pay for is a better decision.

An audit. You want to know whether something meets a standard: your tracking against your consent banner, your checkout against the accessibility rules, your code against basic security practice. A good audit is a measurement. You can check it, and you know what you will do with the result.

A defined decision. Build or buy, which of two suppliers, whether a campaign deserves a bigger budget. The question is narrow, the answer is a choice, and you or your team can act on it the same week.

What these have in common is that the adviser's output is an input to one decision you were going to make anyway. The value of the advice is that it changes the decision, or stops you making the wrong one. Nothing has to be built from it by people who were not in the room.

This is a calculation you can do. The value of a second opinion is the chance that it changes your decision, times what the change saves you, minus the fee. That is expected value: the average result if you could replay the same choice many times, each outcome weighted by its probability.

$$E[X] = \sum_{i} p_i \, x_i$$

Here $E[X]$ is the expected value of the decision, $p_i$ is the probability of outcome $i$, $x_i$ is what you gain or lose if outcome $i$ happens, and $\sum_{i}$ means adding up over every possible outcome. The idea is explained from scratch in Foundations: expected value, and how to use it for business decisions in my article on decision theory for founders.

Worked example: a second opinion that pays for itself

Fionn runs Canice Ceramics in Kilkenny at caniceceramics.example. An agency has quoted him an invented €40,000 to move his shop to a new platform. Before he signs, he pays an independent developer an invented €1,500 for two days to review the quote and his current shop.

Assume, for the example, that there is a 20% chance the review finds the migration is not needed, because his real problem is a slow theme and a broken search that can be fixed for an invented €8,000 on the platform he already has. Then the review saves him $40{,}000 - 8{,}000 = €32{,}000$. There is an 80% chance the review confirms the migration is sensible, and it saves him nothing beyond the reassurance.

Step one, the expected saving: $0.2 \times 32{,}000 + 0.8 \times 0 = €6{,}400$.

Step two, minus the fee: $6{,}400 - 1{,}500 = €4{,}900$.

The review is worth paying for, on these assumptions, even though four times out of five it changes nothing. And notice why it works. The advice feeds one decision Fionn was about to make. He does not need anyone to build anything from it.

Why advice so often fails to turn into growth

Growth and technology rarely look like Fionn's decision. They look like Sorcha's checkout: many small changes, each with snags nobody foresaw, done by people who already have a full week. That is where I think advice alone breaks down, for three reasons.

The first reason is accountability. The adviser is paid for the recommendation, not for the result. If the strategy fails, it is easy to blame the execution. If it succeeds, the adviser can take the credit. I do not say consultants act in bad faith. I say the deal does not ask them to care what happens after the handover, and a deal that does not ask usually does not get.

The second reason is that knowledge does not travel well. The report holds what the adviser could write down. A lot of what makes a change work cannot be written down, or was never noticed until someone tried.

The third reason is the gap between knowing and doing. Even when everybody agrees what should be done, it often does not get done.

No outcome, no accountability

Economists call the relationship between a business owner and anyone acting on their behalf a principal agent problem. The principal (you) wants a result. The agent (the person you pay) controls effort you cannot fully see, and is paid according to the contract, not according to your result. The contract quietly tells the agent what to care about. I go through the theory, with numbers, in my article on principal agent theory and incentives.

Apply that to a consulting engagement. The fee is fixed, or paid by the day. It is the same whether the advice works, half works or sits in a drawer. So what the deal rewards is a report that looks convincing at the handover and an engagement that ends on time. Neither is the same thing as your growth.

There is a second cost, and it is subtler. Someone who is not there when the advice is carried out never finds out whether it worked. Gary Klein, writing about the criteria he and Daniel Kahneman agreed on for real intuitive expertise, names the opportunity for practice with accurate and timely feedback on the consequences of decisions as one of them.¹ An adviser who moves on after the handover loses exactly that feedback. The next client gets advice that was never corrected by the last one.

An example: the consultant writes in Sorcha's report that guest checkout will lift completed orders. Tadhg builds it three months later, after the consultant has moved on. Orders barely move, because the real leak was a delivery charge that only appears on the final step. Nobody tells the consultant. The next report, for the next shop, says the same thing with the same confidence.

Knowledge does not travel in a PDF

When you hand someone a report, you assume the knowledge in it transfers with it. Research on how practices spread inside companies suggests that is optimistic.

Gabriel Szulanski studied 122 transfers of best practice inside eight companies, 271 observations in all. The conventional view blamed motivation: people resist change. He found that the major barriers were about knowledge instead. The receiving team lacked what researchers call absorptive capacity, the ability to take in and use new knowledge. There was causal ambiguity, meaning nobody could say exactly why the practice worked where it came from. And the relationship between the source and the recipient was arduous.²

That study looked at transfers inside one company, where the source is down the corridor and can be asked again. A report from an outside adviser who has moved on faces all three barriers and has fewer ways round them. The recommendations arrive without the reasons behind them, to a team that was not part of the work, from someone who is no longer around to explain.

Worked example: one recommendation, three snags

Take one line from Sorcha's report: "Fix the consent setup so Google Ads receives every conversion it is allowed to receive." It reads like a morning's job. Here is what Tadhg actually runs into.

Snag one, absorptive capacity. Tadhg has never configured consent signals for an ad platform. He spends a day reading documentation before he changes anything.

Snag two, causal ambiguity. The report does not say why conversions are missing. It turns out half the gap comes from the consent banner and half from orders paid through a payment provider whose confirmation page never loads the tag. The report treated it as one problem.

Snag three, the relationship. Tadhg emails the consultant with a question. The engagement has ended, so the answer is a polite offer of an extra day at the day rate.

Assume, for the example, that Tadhg spends 30 hours on this one line at an invented internal cost of €50 an hour. That is $30 \times 50 = €1{,}500$ of his time, for a task the report sized as a morning. Someone who had done it many times would have known about the payment provider before starting. That knowledge is exactly what did not make it into the PDF.

The gap between knowing and doing

Jeffrey Pfeffer and Robert Sutton, both at Stanford, wrote a whole book on this, The Knowing Doing Gap. Its opening question is why there are so many gaps between what firms know they should do and what they actually do. Among the causes they describe is the habit of letting plans and presentations stand in for action.³ A report is a plan. Producing one can feel like progress, for the business as much as for the adviser.

The strongest evidence I know on this comes from a field experiment, and it is fair to say it was designed to test management practices, not consulting as such. Nicholas Bloom and colleagues gave free management consulting to randomly chosen textile plants near Mumbai. Every plant, treated or not, received a month of diagnosis: about 15 consulting days and a written report with recommendations. Only the treated plants then got four months of help putting the practices in place.⁴ The treated plants gained about 17% in productivity within a year, compared with plants that had received the diagnosis alone.⁵

Read that design carefully. The comparison group had the report. What made the difference was the months of hands on work that followed it.

The other strong study points the same way. Miriam Bruhn, Dean Karlan and Antoinette Schoar ran a randomised trial with 432 small and medium businesses in Mexico. The consulting raised productivity and return on assets after a year, and employment and wage bills were about 50% higher five years later.⁶ That is a real result for consulting, and I take it seriously. It was also not a report and a goodbye. The consultants were asked to diagnose the problems, suggest solutions and help the businesses implement them, and they spent four hours a week with each business for a year.⁷ The advice that worked was advice that stayed and helped with the doing.

So my honest reading of the evidence is not that advice is worthless. It is that advice tends to work when the adviser stays close to the doing, over months, with the owner. The less a deal looks like that, the less I would expect from it.

Putting numbers on advice versus the work

The way to compare the two offers is to stop comparing fees and compare expected outcomes. The fee is certain. What you get for it is not. For a consulting engagement, the big uncertainty is whether the recommendations get implemented at all. For a build engagement, that uncertainty mostly moves to the person doing the work.

The expected net value of either engagement is the expected gain, minus the expected cost of your own people, minus the fee.

$$N = \sum_{i} p_i \left( G_i - I_i \right) - F$$

Here $N$ is the expected net value of the engagement, $p_i$ is the probability of outcome $i$ (implemented in full, in part or not at all), $G_i$ is the extra contribution the business earns in outcome $i$, $I_i$ is the internal cost of your own team's time in outcome $i$, $\sum_{i}$ adds up over the outcomes, and $F$ is the fee. It is the same expected value formula as before, with costs subtracted. See Foundations: expected value.

Worked example: Sorcha's two offers, every step

The assumptions first, all invented for the example and stated so you can swap in your own.

If all four changes go live and work, Sorcha's completed orders rise enough to add €1,500 a month of contribution. Over the first 12 months that is $12 \times 1{,}500 = €18{,}000$. If only some go live, say the easy two, the gain is €6,000 over the year.

For the consulting offer, the fee is €6,000 and Tadhg does the implementation. Assume a 30% chance he gets all four done within a few months, at 96 hours of his time, a 40% chance he gets two done, at 48 hours, and a 30% chance the report stays on the shelf because the shop keeps him busy. At €50 an hour his time costs $96 \times 50 = €4{,}800$ in the first case and $48 \times 50 = €2{,}400$ in the second. The reason for the shelf risk: Tadhg already has a full two days a week, and the report adds work without adding days.

For the build offer, the fee is €9,000 and the builder does the work. Assume a 90% chance all four changes go live and work, and a 10% chance only two do, because one change turns out to need the payment provider's cooperation and stalls. Tadhg's time is not needed beyond answering questions, so the internal cost is set at zero for both outcomes.

Step one, the consulting offer, outcome by outcome:

Full: $G - I = 18{,}000 - 4{,}800 = €13{,}200$.

Partial: $G - I = 6{,}000 - 2{,}400 = €3{,}600$.

Shelf: $G - I = 0 - 0 = €0$.

Step two, weight them by their probabilities: $0.3 \times 13{,}200 + 0.4 \times 3{,}600 + 0.3 \times 0 = 3{,}960 + 1{,}440 + 0 = €5{,}400$.

Step three, minus the fee: $N_{\text{consult}} = 5{,}400 - 6{,}000 = -€600$.

Step four, the build offer: $0.9 \times 18{,}000 + 0.1 \times 6{,}000 = 16{,}200 + 600 = €16{,}800$.

Step five, minus the fee: $N_{\text{build}} = 16{,}800 - 9{,}000 = €7{,}800$.

The offer with the lower fee has the lower expected value, by $7{,}800 - (-600) = €8{,}400$. Over the first year, on these assumptions, the consulting engagement loses Sorcha money, while the build pays for itself and then some.

The horizontal axis shows each offer twice, first its fee and then its expected net value after the fee and Tadhg's time. The vertical axis is euro over the first 12 months. The takeaway: the fee is the number you see on the quote, and it points the wrong way. The consulting offer is €3,000 less expensive to buy and about €8,400 worse to own, because most of its value depends on work nobody is paid to finish.

Worked example: when would the report have paid?

Fairness means asking what would make the consulting offer the right one. Leave out the partial case to keep it simple, and let $q$ be the chance that Tadhg implements everything.

The report breaks even when $q \times 13{,}200 = 6{,}000$, so $q = 6{,}000 / 13{,}200 \approx 0.45$.

If Sorcha had a developer with free time, who had done this kind of work before, and she was confident of a better than 45% chance of everything going live, the report could be worth buying. Notice the two conditions behind that number: someone with the skill, and someone with the time. If you have both, advice plus your own team is a perfectly sound deal. Sorcha has neither to spare, and that is the situation this article is about.

Choosing between advice and a build

You can sort most requests with a few questions, asked in order.

Worked example: three requests through the questions

Fionn's question, whether to migrate, is one narrow decision. Advice, priced for the decision, which is what he bought.

Sorcha's checkout and tracking are a change in the business. Tadhg has the skill for some of it but not the time. So it goes to a build. Can the result be measured honestly? Not yet, because the tracking is the thing that is broken. So the first job goes on a fixed quote, and any pay tied to results waits until the numbers can be trusted.

A third case, invented: Rónán runs a cycling parts shop in Limerick with an in house developer who has spare capacity and has rebuilt checkouts before. He wants an outside view of where his checkout leaks. That is advice plus his own team, and it is a good deal for him, because both conditions from the break even example are met.

If you are choosing a person or a firm for the work rather than the type of deal, I wrote a separate guide to choosing a growth agency or specialist.

How I work instead

Here is what I do instead of consulting, using only what my pricing page and my terms say.

I do the work, from setting up campaigns to building automation to fixing analytics. I take responsibility for outcomes, not just recommendations. I stay involved: I track results, adjust the strategy, and iterate until we find what works. And I shape the strategy to the business in front of me rather than reaching for a template.

It starts with a free workshop. I will admit the joke against myself: the workshop is advice. You and I build your plan together, with solutions, growth ideas, goals, timelines and honest cost estimates, and forecasts with a best case, a worst case and error margins. The written plan is yours to keep either way. If I am not the right person for the job, I say so and point you to whoever is.

What makes it different from a consulting report is what comes next. After the workshop come two weeks of real work, on me: a tracking review, the first quick wins, and you watching how I actually operate. If it fits, we carry on. If not, you walk away with the plan and the work already done, at no cost. The workshop and the two weeks are goodwill offers, with no obligation to continue, and the figures in them are estimates, not promises.

For bounded tasks, I publish hourly rates for ad hoc work, and every task still gets a binding quote with clearly defined deliverables before work starts. That answers the accountability question for work with a clear end: you know what you get and what it costs before I begin.

For online shops, I offer a commission model. There is a monthly base fee as the floor, and a commission on the profit my work adds beyond your natural growth. You pay whichever is higher, never both. The baseline is agreed in writing before the work starts, and your ad accounts, tracking and data are yours from day one. My FAQ puts the condition in one line: when tracking is clean enough to measure my contribution honestly, I am happy to tie part of my income to results. Part, not all. The reasons for that, including why a fair commission deal keeps a base fee, are in my article on commission based growth marketing.

An example: put Sorcha through that path, without any prices, because those are agreed per business. The workshop maps the checkout and the tracking and gives her a written plan she keeps. The two weeks on me start on the tracking review, the job that has to come first because every later number rests on it. The checkout changes then go on a binding quote with the deliverables written down, live on her shop rather than in a report. Only once the tracking is clean would a commission on added profit even be on the table. At no point does she hold a document that someone else still has to turn into a working shop.

What this means for you

If you run an eCommerce shop or a SaaS business with a limited budget and big ambitions, you do not have the luxury of paying for advice that might work.

Buy advice when you need a decision checked, a standard measured, or one question answered, and when your own people have the skill and the time to act on it. When the job is growth, a better checkout, cleaner data, campaigns that pay, buy the work from someone who does it and answers for whether it works.

Ask every offer the same question: when this engagement ends, who is holding the risk that nothing changes? If the answer is you, price that risk in before you compare fees.

Footnotes

1. Gary Klein, Conversations With Kahneman, Psychology Today, 18 April 2024, on the criteria agreed in Daniel Kahneman and Gary Klein, Conditions for Intuitive Expertise: A Failure to Disagree, American Psychologist, volume 64, issue 6, 2009, https://www.psychologytoday.com/gb/blog/seeing-what-others-dont/202404/conversations-with-kahneman

2. Gabriel Szulanski, Exploring Internal Stickiness: Impediments to the Transfer of Best Practice Within the Firm, Strategic Management Journal, volume 17, winter special issue, pages 27 to 43, 1996, https://doi.org/10.1002/smj.4250171105

3. Jeffrey Pfeffer and Robert I. Sutton, The Knowing Doing Gap: How Smart Companies Turn Knowledge into Action, Harvard Business School Press, 1999, book page at Stanford Graduate School of Business, https://www.gsb.stanford.edu/faculty-research/books/knowing-doing-gap-how-smart-companies-turn-knowledge-action

4. Abdul Latif Jameel Poverty Action Lab, Increasing Firm Productivity through Management Consulting Services in India, evaluation summary, Abdul Latif Jameel Poverty Action Lab, https://www.povertyactionlab.org/evaluation/increasing-firm-productivity-through-management-consulting-services-india

5. Nicholas Bloom, Benn Eifert, Aprajit Mahajan, David McKenzie and John Roberts, Does Management Matter? Evidence from India, The Quarterly Journal of Economics, volume 128, issue 1, pages 1 to 51, 2013, https://econpapers.repec.org/RePEc:oup:qjecon:v:128:y:2013:i:1:p:1-51

6. Miriam Bruhn, Dean Karlan and Antoinette Schoar, The Impact of Consulting Services on Small and Medium Enterprises: Evidence from a Randomized Trial in Mexico, Journal of Political Economy, volume 126, issue 2, pages 635 to 687, 2018, https://econpapers.repec.org/RePEc:ucp:jpolec:doi:10.1086/696154

7. Miriam Bruhn, Dean Karlan and Antoinette Schoar, The Impact of Consulting Services on Small and Medium Enterprises: Evidence from a Randomized Trial in Mexico, working paper version, section on the programme design, hosted by GOV.UK research outputs, https://assets.publishing.service.gov.uk/media/57a09dd540f0b64974001a90/sme_consulting_mexico.pdf