Element Group builds and runs AI systems for sales, marketing and operations. Each project starts with a business metric, such as revenue, cost or time, and a plan to measure the result.
Marketing track record: brands we have run search and paid media for
Polaroid
Sony
Alibaba
Xiaomi
Yves Rocher
GE Money Bank
Cisco
Subaru
Nespresso
River Island
PolaroidSonyAlibabaXiaomiYves RocherGE Money BankCiscoSubaruNespressoRiver Island
PolaroidSonyAlibabaXiaomiYves RocherGE Money BankCiscoSubaruNespressoRiver Island
Problem
You have AI tools. Are they improving the business?
Pilots stall. Data stays fragmented. Routine work takes as long as it did. These are the gaps we look for before proposing a build.
1
Pilots that never reach production.
No clear owner, integration plan or success metric.
2
Tools that don’t share data.
Sales, advertising and finance data sit in separate systems.
3
Your team doing a machine’s job.
Call notes and reports are still written by hand.
4
Results nobody can measure.
No reliable before and after for cost, time or business outcomes.
AI projects
AI in production. Here’s what we measured.
Three systems we built and still run inside our own group’s businesses: an ad audit finding, a cut in call write-up time and content output. Every figure is our own measurement.
Internal project · Consumer electronics · Three marketplaces
29%
no salesaudited ad spend
of audited ad spend went to campaigns with no sales recorded in the audited data.
Data lived in three marketplace accounts, accounting, the bank and spreadsheets. Leadership saw revenue but not which product was losing money. Now daily syncs feed unit economics per product, and a morning ad audit flags campaigns for review. Changes to live ad accounts wait for human approval.
Internal project · B2B sales
Call write-up time went from about 60 minutes to 5605min
5 min now60 min before
to write up a sales call, before and after. Every call is now analysed, not a sample.
Every call is transcribed and summarised straight into the deal history, with coaching notes for the manager and a team digest of objections: the reasons buyers give for saying no, and how often each comes up. Script changes now start from that digest.
Internal project · B2B services website
250
58 blog192 wiki
articles produced by two AI content pipelines: 58 for a commercial blog, 192 for a reference wiki.
Topics come from live search demand. Agents draft, facts are checked against primary sources, and nothing is published until it passes a multi-layer quality gate. Experts review every draft.
If it can’t tell a revenue story, we don’t ship it.
What we build
Six ways we put AI to work.
Each system plugs into a workflow you already run, and we agree up front how its impact will be measured.
Sales
Sales call intelligence
Sales calls turned into CRM notes, coaching points and a running count of the objections buyers raise.
Advertising
Ad spend monitoring
Campaign spend and sales data in one place. Campaigns are flagged for review before any budget changes.
Content
AI content workflows
From topic research to reviewed drafts, with facts checked against primary sources before publication.
Leadership
Business performance dashboards
Unit economics and channel performance in one view, with data updated automatically.
Customer service
Customer response drafts
Replies to reviews and questions drafted in your brand’s voice. A person approves each one before it is sent.
Team
AI tools your team can run
Agents and playbooks your own people use in their own workflows. You get the method as well as the system.
Start with one process and a clear success metric.
We begin with an audit and a scoped pilot. The target for a first pilot is four weeks; the actual timing depends on the scope and the systems involved.
1
Define the business case
Choose one workflow, record the baseline, and agree the success metric and acceptance criteria.
2
Build and test
Build the agreed workflow, test the integrations and review the result with your team.
3
Measure and decide
Compare the result with the baseline, then decide whether to expand, revise or stop.
We start with an audit and a clear business case. If automation is a poor fit, we’ll say so.
Phased delivery
A written scope with acceptance criteria. You sign off before development starts.
Code and security reviews
Code is reviewed at every phase. Work that touches financial actions, account access or personal data also gets a security review.
Result checks
Each automation checks its result, not just that it ran. Empty output counts as a failure.
Approval before money moves
Actions with financial impact need explicit approval, and every one is logged.
Data protection
In the systems we build, personal data is anonymised in-house before it reaches a third-party model.
The people on your project
A strategist, a marketer, an AI engineer and an analyst. Together they define the business case, build the workflow and check the numbers before and after.
Marketing track record
Two decades of marketing results for brands you know.
Before AI, our work was search, performance and product marketing, including for the brands below. It predates our AI practice and shapes which sales and marketing problems we choose to solve first.
Selected marketing work and results
Brand
Sector
Work
Result
Polaroid
Cameras
Paid search and content
4.2× revenue from paid search
Sony
Online store
Search, paid and display mix
1.9× sales from paid traffic
Alibaba
E-commerce
SEO strategy and technical requirements for 1M+ pages
+386% organic traffic
Xiaomi
Consumer electronics
Search visibility for the US site, Google and Yahoo
+378% organic traffic
Yves Rocher
Beauty
Organic search
+193% organic traffic
GE Money Bank
Banking
Organic search
+345% site visits
Cisco
Web conferencing software
Search marketing programme on Baidu
+87% site traffic
Subaru
Automotive
New model launch campaign
3.5× average CTR
Nespresso
Coffee
Organic search for the online store
31% increase in queries ranking no. 1
River Island
Fashion
SEO across 3,000 pages
4,108 keywords ranking in the top 10
Selected marketing engagements completed before our AI practice. Results are drawn from our own case studies. All trademarks belong to their owners; no endorsement is implied.
Where our teams are based, with the local time in each country.
Tell us where the money leaks.
Tell us which process you want to improve, which systems it uses and what you want to measure. We’ll tell you whether AI is a good fit and what a pilot could involve. If it isn’t, we’ll say so.