The GROW Methodology™ — profitable growth in Google Shopping

The GROW Methodology

A tried and tested framework for profitable, sustainable sales growth in Google Shopping. Refined over twenty years in the e-commerce industry, and responsible for more than £1.2 billion in additional revenue for retail brands.

  1. 01Track every cost
  2. 02Calculate precision bids
  3. 03Build for reach

Introduction

Why we published this

We have spent twenty years inside e-commerce advertising accounts. Every year of it has gone back into the method: testing what holds, discarding what does not, and tuning it across thousands of them until it became a repeatable process rather than a collection of tactics.

In that time we have managed more than 72 million products across 31 countries and delivered over £1.2 billion in additional revenue. What follows is the method we use to do it.

This is not the summary version. It is the actual method, with the arithmetic shown. The GROW Methodology is a long way from what most Shopping accounts are doing: it contains no magic buttons and very few shortcuts, and in several respects it looks like hard work, because it is. It draws on commercial strategy, financial analysis, statistics, data engineering and merchandising judgement, and it asks for intelligence, experience and craft across a wide range of disciplines. Experience has shown that it works, repeatedly, across categories, catalogue sizes and markets. It is what we do every day for the brands we run.

If you are spending real money in Google Shopping and cannot presently say which of your products make a profit after advertising, the next section is the one to read.

The problem

Why profitable brands lose money in Shopping

It is almost never bad management. It is an arithmetic error that no standard report is built to show you.

For most e-commerce brands, paid search is among the largest controllable costs in the business and the least understood by the people who sign it off. Media costs rise each year. Competition increases. Suppliers pass on their own inflation while price pressure at the shelf takes what is left. Margin is compressed from both ends at once, and the paid search team is asked to deliver growth regardless.

What happens next inside the account is easy to describe, and we have now watched it happen several hundred times.

Campaigns get organised the way a merchandiser thinks about a range: by brand, by category, by product type. It is a sensible way to organise a catalogue and a poor way to bid on one. Fifty products end up sharing a product group. Sometimes five hundred. Each of them carries a different cost price, a different delivery cost, a different return rate and a different margin, so each of them needs a different return on ad spend to produce the profit the business requires.

The group can only carry one target, so somebody sets an average. To see why that is a problem, look at the shape of a real catalogue.

Figure 1

Gross margin across a 3,981-SKU homewares catalogue

310
520
690
840
720
480
290
131
<10%10–20%20–30%30–40% 40–50%50–60%60–70%70%+

Gross margin band · SKU count

The band containing the catalogue average (36%) Every other product
The weighted average margin is 36%, which is what a blended target ends up being built on. Only 840 of 3,981 products — 21% of the catalogue — actually sit in that band. The average describes a fifth of the range and misprices the rest.

Now apply the arithmetic. The return on ad spend a product needs simply to break even is one divided by its gross margin. That is not a straight line. It is a curve, and it is a steep one at the bottom.

Figure 2

Break-even ROAS against gross margin, with one blended target applied across the catalogue

15% margin needs 6.7×, gets 2.8× 65% margin needs 1.5×, gets 2.8× Correct at 36% only Overspending Underbidding 0× 2× 4× 6× 8× 10× 10% 20% 30% 40% 50% 60% 70% 80% Gross margin Break-even ROAS Blended target 2.8×
Above the curve a product returns more than it costs; below it, every sale takes money out of the business. A single horizontal target crosses the curve at exactly one point. To the left of it the account is buying sales it cannot afford. To the right it is refusing sales it could have won at a profit.

Both halves of that chart cost you money, and they cost you money in ways that look nothing alike. On the left the damage is visible eventually, in the P&L. On the right it is never visible at all, because you cannot report on the orders you did not take.

Neither error shows in the ROAS column of the account, for the plain reason that the column is an average too. A group returning 4× might be a healthy product carrying a bad one, and nothing in the interface will tell you which.

“Cutting the budget and raising the target are the only two levers most accounts have. Both trade volume for the appearance of efficiency.”

The loss begins on day one. It becomes visible at month end, or in many cases at quarter end, when finance reconciles ad spend against realised sales and margin. On a brand spending £150,000 a month, that is close to half a million pounds committed before anybody can prove what it earned.

The correction that follows is always the same, and it is the reason we call this a cycle rather than a mistake.

Figure 3

The correction cycle

  1. 01An average target is setOne ROAS across a group of products with entirely different margins.
  2. 02Profit leaks, unseenThin lines overspend, strong lines are priced out. Daily reporting shows neither.
  3. 03Finance finds the gapSpend is reconciled against realised margin weeks after the money left.
  4. 04The account is correctedBudgets cut or targets raised. Volume drops, so the targets are loosened again.
and it starts again
Each correction is rational on its own terms and wrong in aggregate. Some brands have run this loop for years, arriving back at the same blended target every eighteen months.

So the diagnosis is not laziness or inexperience. It is that the account is being measured at the wrong resolution and at the wrong speed: an average bid, judged by an average figure, weeks after the money left the bank.

Change the resolution and the speed and the loop stops. That is all the method does. It just has to do it for every product in the catalogue, every day, which is why it took twenty years and a good deal of software to make practical.

Our 3 core principles

They run in a fixed order, because each one depends on the answer produced by the one before it. Taken out of sequence the method does not work: a precise bid calculated from the wrong cost is precisely wrong.

  1. 01
    Track every costKnow the exact margin on every product before you bid on it.
  2. 02
    Calculate precision bidsSolve each product's target from the profit you want to keep.
  3. 03
    Build for reachGive every product a structure that can carry its own bid.
01

Track every cost

Fully costed ads — MarginStackTM

If you cannot currently say what a single sale of a single product is worth to you after everything has been paid, start here. Nothing further down this page will work until you can.

Every business runs on the same arithmetic: what a thing costs, what it returns, and the gap between the two. Google Shopping is one of the few places in an e-commerce operation where that arithmetic is not done. Bids are placed against revenue, and revenue is not profit.

The distinction matters more than most teams expect, because the better-looking campaign is very often the one draining the business. Two products, the same £100 of revenue each:

Figure 4

Where £100 of revenue actually goes

Product A · 15% gross margin · bidding at 5.0× ROAS
£85.00
£20.00
£100 revenue

Costs exceed revenue by £5.00 on every sale

Product B · 60% gross margin · bidding at 2.5× ROAS
£40.00
£40.00
£20.00
£100 revenue

£20.00 of profit banked on every sale

Cost of the sale Ad spend Profit Beyond the revenue
Product A returns twice the ROAS of Product B and loses money on every order. In every standard report, Product A is the campaign that looks healthy — and it is usually the one that gets more budget.

MarginStackTM is how we close that gap. It holds the full cost stack behind a single sale: cost of goods, delivery, packaging, payment fees, returns, import duties, customer service, platform and channel fees. It then adds back what the sale is genuinely worth, including measured post-sale revenue and lifetime value from repeat buyers. Fifteen or more cost lines, held against the individual SKU rather than averaged across the range it happens to sit in.

Two of those lines are the ones teams most often leave out, and they are usually the two that decide the answer. Returns are a percentage of revenue that never becomes revenue, and they vary enormously by product — apparel and furniture behave nothing like consumables. Lifetime value works the other way: a product that reliably brings a customer back is worth bidding harder on than its first-order margin suggests. Leave both out and you will underprice your best products and overprice your worst.

How to implement it

  1. 1
    Pull the cost lines you already holdCost of goods and delivery live in your ERP or store. Payment fees are on your processor statements. Import duties and packaging are usually in a finance spreadsheet.
  2. 2
    Measure returns per product, not per accountAn account-wide return rate is the same averaging error one level up. Use at least twelve months so seasonal lines are represented.
  3. 3
    Add post-sale revenue and lifetime valueMeasure it, do not assume it. Repeat rate multiplied by average repeat order value, attributed to the product that acquired the customer.
  4. 4
    Keep it currentCosts move. A supplier increase or a shipping-rate change that is not reflected within days puts every downstream bid out by the same amount.

In GROW this is a single import from a sheet, a CSV or your store, after which the agents maintain it. Doing it by hand is entirely possible for a few hundred SKUs and unmanageable beyond that, which is the recurring theme of this page.

Outcome

You know the profit on every product, every day. Nothing that follows can work without it.

02

Calculate precision bids

Profit control — ProfitClarityTM

Once true costs are known, the bid stops being a judgement and becomes a calculation. This is where the money is.

Decide the profit you want the account to return, as a percentage of revenue. Everything else follows from it, because the ad spend a product can absorb is simply its gross margin less the profit you intend to keep — and the target ROAS is the inverse of that number.

Written out, the whole of principle two is one line:

Figure 5

Solving each product's target from a single 12% profit goal

Target ROAS = 1 ÷ (gross margin − profit target)
Gross margin Ad spend allowed
per £100 sale		 Break-even
ROAS		 Target ROAS
at 12% profit
15%£3.006.67×33.3×
20%£8.005.00×12.5×
30%£18.003.33×5.56×
40%£28.002.50×3.57×
50%£38.002.00×2.63×
60%£48.001.67×2.08×
70%£58.001.43×1.72×
The same profit goal produces targets ranging from 1.72× to 12.5× across one catalogue. The 15% row is the useful one: at a 12% goal that product can absorb £3.00 of advertising and would need 33× to do it — which is the method saying, in arithmetic, do not advertise this product.

That last row is worth dwelling on, because it is the answer a blended target can never give you. A single account-level number quietly funds products that cannot carry advertising at all. Solving per product tells you which ones they are, and it tells you before the money is spent rather than at the quarterly review.

It also tells you when to break your own rule. Where measured lifetime value or post-sale revenue is strong, bidding below break-even on the first order is a sound investment — you are buying a customer, not a sale. The difference between that and the accidental version is that here it is a decision, with a number attached, made deliberately.

ProfitClarityTM does this calculation for every SKU in the catalogue at the same moment and writes the answers into Google Ads. Move the profit target and the whole account is re-solved and synced inside a minute. Raise it ahead of a tight quarter and spend concentrates on the products that can carry it. Lower it during a growth push and the account buys reach on purpose, product by product, with the cost of that decision known in advance rather than discovered in arrears.

How to implement it

  1. 1
    Agree the profit target with finance, not with marketingIt is a business decision about what the account must contribute. Marketing's job is to hit it, not to set it.
  2. 2
    Solve the target for every SKUOne divided by (margin − target). Do it per product; a category-level version reintroduces the error you just removed.
  3. 3
    Flag the products that cannot make itAnything needing an implausible ROAS is telling you to fix the product economics or stop advertising it. Both are valid; ignoring it is not.
  4. 4
    Re-solve on every changeA price change, a supplier increase, a shift in return rate or a new profit goal all move the answer. Stale targets are the same problem as no targets.
Outcome

You set the profit the business needs, and every bid in the account is held to it.

03

Build for reach

Campaign structure — one SKU, one group, one bid

This is the principle most teams skip, and it is the one that produces most of the growth.

A precise bid is worth nothing if the campaign structure cannot carry it. Put fifty products back under a single product group and they are forced back onto a single number. The work done in the first two principles is thrown away at the last step, and the account performs exactly as it did before.

So every nurtured product is given its own product group and its own target.

Figure 6

The same 212 products, structured two ways

Conventional structure

Cookware 212 products in one product group tROAS 4.0×
4.0×4.0×4.0×4.0×4.0×4.0×4.0×4.0×4.0×4.0×4.0×4.0×4.0×4.0×4.0×4.0×4.0×+195

Margins in this group run from 9% to 63%. One target cannot be right for that spread, and Google receives one blended signal for all 212.

GROW structure

Cookware 212 product groups, one per SKU 212 distinct targets
2.1×5.6×3.4×2.6×8.3×3.1×2.2×4.4×1.9×6.7×2.9×3.8×2.4×5.0×2.7×3.3×4.8×+195

Each product bids at the level its own margin supports, and Google's bidding receives a clean signal per product — the condition it was designed to work under.

The structure on the right is not a clever tactic. It is the only structure in which a per-product bid can survive contact with the auction.

Three things change once the structure is in place. Strong products scale without being held back by weaker neighbours. Weak products throttle without dragging a whole category down with them. And nothing subsidises anything else, so the numbers you read are the numbers.

Reach follows from this directly, and it is where most of the additional revenue comes from. A product that a group average had priced out of the auction now bids at the level its own margin supports, so it starts appearing on searches it never used to reach. The account is not spending more. It is spending accurately, in more places.

The structure is neither new nor secret. It is simply impossible to build and maintain by hand at any real catalogue size, which is why very few accounts are built this way, and why we said at the top of this page that most brands will not implement the method themselves. Two hundred product groups is a long afternoon. Four thousand, kept current as the range changes, is not a job a person can hold.

How to implement it

  1. 1
    Decide which products are worth nurturingNot everything earns its own group. Start with the products carrying meaningful revenue or meaningful margin, and let the long tail sit in a catch-all until it proves itself.
  2. 2
    Split down to one product per groupUse item ID as the final subdivision so each SKU can hold its own target. Everything above it is organisation, not bidding.
  3. 3
    Apply the solved target from principle twoEach group gets the number calculated for that product, not for its category.
  4. 4
    Automate the intakeNew lines, seasonal ranges and end-of-line stock arrive constantly. If placing them is a manual task, the structure decays within a quarter.
Outcome

The largest single lever in the method, and the one that automation finally puts within reach.

Running it

The method in operation

Principles only pay on the day the money is being spent. Four things make that possible.

Delivered with

Speed

A method is worth what it is worth on the day it is applied. Campaigns are built in minutes rather than weeks, and bids move within moments of a price change, a supplier increase or a shift in return rates. The account is never trading on last month's costs.

  • Campaigns built, costed and live in minutes
  • Bids recalculated the moment a cost, price or target changes
  • Structures that no team could maintain by hand, maintained continuously

Planned for

Cashflow

Ad spend leaves the business weeks before the margin it earns arrives. That gap is what makes an average bid dangerous during a sale or a scale-up: volume rises, cash goes out faster, and the profit assumed to be behind it may not be there. Every bid is costed before it is placed, so the cash consequence is known when the decision is made.

  • Every bid costed in advance, not reconciled afterwards
  • Discount modelling — profit per sale at 5%, 10% and 15% off, and the break-even price
  • Budget pacing tracked daily: on pace, over or under

Measured with

Sale impact

A promotion that lifts revenue can still lose money, and the report showing the lift rarely shows the loss. Every sale period is measured against a like-for-like baseline: revenue, profit after the discount, return on ad spend, and the lift that holds in the weeks afterwards.

  • Profit after discount, not revenue alone
  • Post-sale lift tracked for weeks after the promotion closes
  • Produced automatically, with no spreadsheet to build

Measured with

Change impact

Most changes made to an advertising account are never measured, so accounts fill up with decisions nobody can explain. Every change is recorded on the timeline, whether it was a price, a cost, a target or a whole campaign, and whether a person or an agent made it. Beside it sits what happened next.

  • Who changed what, when, and what it did
  • Before and after on the metric you choose
  • A record you can take into a board meeting and defend

Putting it in place

Where to start

The sequence matters more than the speed. This is the order we use on every account we take on.

Figure 7

Implementation sequence

  1. Day 1ConnectGoogle Ads and Merchant Center linked, the catalogue read in full.
  2. Day 1CostCost lines loaded per SKU and reconciled against finance.
  3. Day 2CalculateProfit target agreed; a target ROAS solved for every product.
  4. Week 1StructureCampaigns rebuilt so each nurtured SKU carries its own bid.
  5. OngoingMeasureEvery change and every promotion measured against baseline.
Steps one to three are arithmetic and can be done in a spreadsheet for a small catalogue. Step four is where scale defeats manual work, and step five is where most accounts quietly stop.

If you would rather not build it yourself, that is what we do. Our agents run the whole sequence on your account, and the setup is done for you. But the method is the method whether we run it or you do, and we would sooner see it used than owned.

Honest limits

What it is not

  1. 1
    It is not a bidding scriptScripts react to yesterday's performance. The method starts from cost data and calculates the correct bid before the money is spent.
  2. 2
    It is not a replacement for your teamAgents take the repetitive, analytical work: costing, calculating, structuring and checking. Range, pricing, brand and strategy stay where they belong.
  3. 3
    It is not a quick winIt is a repeatable process. The gains come from applying it to every product, every day, and from what each measured change teaches the account.
  4. 4
    It will not rescue a product that does not workIf the economics of a line are broken, the method will tell you clearly and early. It will not make the line profitable.

Results

What it produces

Clarity, consistency and control over marketing spend — and they arrive in that order.

Clarity comes first: you know the profit on every product, and you know it today rather than at the end of the quarter. Consistency follows, because the same calculation is applied to every SKU, every day, whether the catalogue holds four hundred products or forty thousand, and it does not vary with who happens to be on shift. Control is the result: you set the profit the business needs and the account is held to it.

The pattern across the brands running the method is consistent. Margin improves by roughly 14% in relative terms, because spend stops flowing to products that cannot carry it. Sales grow by between 15% and 450%, depending on the size of the brand and the stage it has reached, because products that were priced out of the auction start appearing in it.

The two move together, which is the whole point. Most accounts can only deliver one at the expense of the other, and that trade is the thing brands come to us to stop making.

14%relative improvement in profit margin
15–450%growth in sales, by brand size and stage
72m+products managed under the method

See how you measure up

Answer five questions and get a scorecard against each principle: where you are strong, where profit is leaking, and what GROW would do about it.

Find your growth