Tools and providers
Part of A procurement process for AI marketing tools
Why an AI marketing tools supplier comparison fails without weightings
Compare AI marketing tools suppliers on scored criteria, data protection duties, procurement evidence and total cost before you sign a contract.
What to take away
- The guide to using artificial intelligence in the public sector, published in 2019, treats AI adoption as staged decisions with named owners rather than one purchase. That structure suits marketing teams comparing suppliers.
- Score every supplier on one weighted sheet, because demonstrations flatter different strengths and hide gaps.
- Ask data protection questions in the first round, before a contract draft exists.
- Compare twelve-month total cost, including seats, overages and internal time.
- Keep the scoring notes. They form your audit trail when a decision is challenged.
A supplier comparison for AI marketing tools is paperwork with a commercial edge. The goal is a defensible shortlist, not a favourite.
Which criteria separate AI marketing tools suppliers?
Most comparisons collapse because each supplier is judged on the criteria it happens to do well. Fixing the criteria first solves that.
Six criteria to fix first
- Problem fitthe job and its owner
- Data handlinglocation, retention, access
- Model transparencynamed models, change frequency
- IntegrationsCRM, advertising, analytics without middleware
- Supportnamed contact, response times, escalation
- Commercial termsnotice, seats, overage, exit
Weight the list before you score anything. A small in-house team might give integrations and support 40% between them, while a regulated brand pushes data handling higher. Note the weights and the date, because memory rewrites them later.
The AI marketing tools supplier guide for 2027 maps the wider provider field, so read it before you decide how many suppliers to approach.
How should you weight data protection in the comparison?
Data protection is a scoring line, not a disclaimer. Buyers need the lawful basis for processing, the retention period, and whether customer content trains a shared model.
The ICO's guidance on AI and data protection explains how data protection law applies across an AI system's lifecycle, including fairness and transparency. Ask which duties the supplier carries and which sit with you.
Watch for vague answers about sub-processors, training data taken from customer content by default, and transfer routes outside the UK that nobody can name.
What does procurement evidence look like for public sector buyers?
Public sector marketing teams face more scrutiny. The GOV.UK guide separates problem definition from testing and live monitoring, so supplier evidence should follow the same stages.
Ask for a DPIA template, a security certification, a sub-processor list, two named references and a written exit plan. Private sector buyers get the same benefit at contract review.
How do you compare total cost rather than headline price?
Headline licence fees hide seat minimums, usage tiers, model surcharges and setup time. Build a twelve-month view with the same assumptions for every supplier. For example, six seats at £400 a month is £28,800 a year, and a usage-based quote needs converting on the same basis.
Where a supplier touches media buying, check how it fits programmatic workflows. The IAB UK's Back to Basics guide to programmatic explains the trading models behind media costs, which shows where a tool adds fees to the chain.
Add internal cost too: data preparation, training days and the analyst time spent checking outputs in the first quarter. A 40-hour setup at an internal rate of £60 an hour adds £2,400 before the tool does any work.
Once the shortlist is signed off, the harder task is sequencing rollout. Our guide to AI marketing tools implementation in England covers that next stage in detail.
Common questions
How many suppliers should I score?
Three to five is workable. Fewer gives no contrast, and more than six makes the sheet shallow.
Do I need a DPIA before running a trial?
If personal data is involved, treat it as required. The trial results are more useful when the assessment already exists.
Can I score a free trial and a paid pilot on the same sheet?
Yes, provided the criteria and weights are identical. Note any difference in data access or support, or the results are not comparable.
What happens when two suppliers tie?
Return to the weights you wrote down. If the tie survives that, compare exit terms, because leaving a tool is harder than joining one.
