5 Signs Your E-Commerce Tech Stack Is Bleeding Margin

The Problem Nobody Wants to Audit

I’ve spent the last few years pulling apart e-commerce tech stacks. Fashion retailers, marketplace platforms, DTC brands doing £10M–£200M in revenue. And the pattern is almost always the same: the tech bill has grown 3–5x since launch, nobody can explain exactly why, and the CFO is starting to ask uncomfortable questions.

The uncomfortable answer is usually that 20–40% of tech spend is waste. Not incompetence — just accumulation. Tools get added during a crisis and never removed. Contracts auto-renew. Infrastructure scales up but never scales back down. And because tech is treated as a black box by leadership, nobody connects the dots between what the business is paying and what it’s actually getting.

Here are the five signs I see most often. If three or more apply to you, your ecommerce tech stack costs are probably eating margin you could recover in weeks, not months.

1. You’re Paying for Overlapping Tools That Do the Same Thing

This is the single most common problem. It’s also the easiest to fix.

I recently audited a fashion e-commerce company running three separate email platforms. Klaviyo for marketing automation. Mailgun for transactional emails. And a legacy SendGrid account that nobody could remember setting up, still sending 40,000 order confirmation emails a month at £800/month. Two of those three were doing the same job.

But email is just the obvious one. The real waste hides in analytics. I regularly find companies running Google Analytics, a paid analytics tool like Mixpanel or Amplitude, a heatmap tool like Hotjar, a BI platform like Looker, and then a custom dashboard someone built in Metabase two years ago. Five tools. Three of them overlap significantly. Two of them have zero active users in the last 90 days.

This happens because tool purchases are decentralised. Marketing buys their stack, engineering buys theirs, the data team buys theirs, and nobody has a single view of what the business is actually paying for. Each tool made sense in isolation, at the time. But nobody ever goes back and consolidates.

What to do about it: Run a tool audit. Not a spreadsheet exercise — actually pull login data and API usage for every SaaS subscription. If a tool hasn’t been meaningfully used in 60 days, cancel it. If two tools overlap, pick one and migrate. This alone typically saves 10–15% of total SaaS spend.

2. Your Cloud Bill Hasn’t Been Audited in 12+ Months

Cloud infrastructure bills are designed to be confusing. AWS, GCP, and Azure all have pricing models complex enough that entire companies exist just to help you understand your bill. That complexity is not accidental.

If you haven’t done a proper cloud audit in the last year, you are almost certainly overpaying. I don’t mean glancing at the monthly total. I mean a workload-by-workload review: what is each service actually doing, what resources does it actually consume, and is the current provider the right one for that specific workload?

We did exactly this for a client spending £38,000 per month on AWS. After auditing every workload, we migrated the commodity compute to Hetzner and kept only the genuinely AWS-dependent services on AWS. The result was a 60% reduction in hosting costs — £276K in annualized savings. And the migration was completed with zero downtime.

The most common findings: staging environments running on production-grade instances, auto-scaling groups that only ever scale up, reserved instances that expired months ago and reverted to on-demand pricing, and storage volumes attached to instances that were terminated six months ago.

What to do about it: Schedule a cloud audit. Not next quarter — this month. Tag every resource, match it to a workload, and compare what you’re paying to what the workload actually needs. If your team doesn’t have time, that’s exactly what we do in our first two weeks of an engagement.

Curious what your margin opportunity looks like?

3. You’re Paying Enterprise SaaS Prices for Features You Don’t Use

SaaS vendors are very good at selling you the enterprise tier. They’ll demo the advanced features, show you the roadmap, and convince you that you’ll need those features “as you scale.” Then three years later, you’re paying £60K/year for a platform and using maybe 20% of it.

I see this most often with search and personalisation platforms, CRM systems, and marketing automation tools. A client was paying £72K/year for a third-party search and merchandising platform. When we looked at actual usage, they were using basic keyword search and a handful of merchandising rules. The AI-powered personalisation features they were paying a premium for? Never configured.

We replaced that platform with an in-house solution using open-source search and a lightweight AI layer. The result was better search relevance, full ownership of the technology, and £65K/year in savings.

This isn’t always about building in-house. Sometimes it’s about downgrading to the tier you actually use, or switching to a competitor that prices more sensibly for your usage pattern. The point is that you should be paying for what you use, not for what a sales rep convinced you that you might use.

What to do about it: For every SaaS tool costing more than £10K/year, list the features you actually use versus the features you’re paying for. If you’re using less than 40% of what you’re paying for, renegotiate or replace.

4. Your Database Is Running Unoptimised Queries That Burn Compute

This one is more technical, but the financial impact is often the largest. Bad database queries don’t just make your site slow — they burn compute. And compute costs money.

On a recent engagement, we found a single product catalogue query that was executing 14,000 times per day, each time doing a full table scan across 2.3 million rows. The query took 8–12 seconds per execution. The database server was scaled to a £2,400/month instance just to handle the load from this one query and a handful of others like it.

After rewriting the query and adding proper indexing, execution time dropped from 8 seconds to 0.00014 seconds. A 59,000x improvement. The database server could then be downsized to an instance costing £600/month. One query fix saved £21,600/year in compute costs alone.

This is shockingly common in e-commerce platforms, especially those built on Magento, WooCommerce, or custom frameworks that have been extended over several years. Every plugin, every custom feature, every integration adds queries. And nobody ever goes back to check whether those queries are efficient.

What to do about it: Enable slow query logging. Identify your top 20 most expensive queries by total execution time (frequency multiplied by duration). Fix the worst five. You’ll likely be able to downsize your database infrastructure as a result. If your team doesn’t have the capacity, this is a two-week exercise we do routinely.

5. You Have No Idea What Your Actual Cost-Per-Order Is

Ask most e-commerce CEOs what their cost-per-order is and they’ll give you a number that includes fulfilment, shipping, and payment processing. Ask them what it costs in tech overhead to process that order — the infrastructure, the SaaS tools, the integrations, the engineering time — and you’ll get a blank stare.

This is the root cause of all the other problems on this list. If you don’t know what tech costs per order, you can’t make rational decisions about tech spend. You can’t prioritise which costs to cut. You can’t evaluate whether a new tool will improve margin or erode it.

I worked with a DTC brand doing 400,000 orders per year with a total tech spend of £840K. That’s £2.10 per order in tech overhead. Their average order value was £45, and their gross margin was 62%. So tech was consuming 7.5% of their gross margin — more than their payment processing fees. Nobody in the business knew this number before we calculated it.

Once you have that number, decisions become clearer. A £30K/year tool that doesn’t measurably improve conversion or reduce costs elsewhere is adding £0.075 to every order. That sounds small until you realise it’s compounding across dozens of tools.

What to do about it: Calculate your total tech spend (infrastructure + SaaS + engineering allocation + agency/contractor costs). Divide by annual order volume. Compare that number to your gross margin per order. If tech overhead is above 5% of gross margin and you’re not seeing clear ROI, you have a problem worth solving.

This Is a Leadership Problem, Not a Tech Problem

I want to be clear about something: this is not about blaming your engineering team. In almost every case I’ve seen, the tech team knows where the waste is. They just don’t have the mandate, the time, or the business context to fix it. They’re busy shipping features, fighting fires, and keeping the lights on.

The problem is that leadership — the CEO, the CFO, the board — treats tech spend as a single line item. They see a number going up and they either accept it or demand a blanket cut. Neither approach works. What works is a structured audit that maps every pound of tech spend to a business outcome.

That’s what we do at MarginOps. We work alongside your CTO and your engineering team, not instead of them. We bring a financial lens to technical decisions and a technical lens to financial ones. Our typical engagement starts with a 2–3 week rapid assessment: we map the full tech stack, identify the waste, quantify the opportunity, and build a prioritised roadmap.

Then we execute alongside the team. We don’t hand over a PDF and disappear. We help ship the changes, upskill your engineers on the optimisation patterns we find, and make sure the savings actually materialise. Most clients see the first savings within 30 days.

The goal is to reduce ecommerce operating costs without reducing capability. Cut the waste, keep the value, and give leadership a clear picture of where every pound goes.