Systems that do the follow-up.

Four things that only work together. It fixes the data foundation, because every automated decision inherits whatever your records believe. It designs and builds the journeys — onboarding, nurture, recovery, renewal, winback — on paper first and in the platform second. It writes the content those journeys carry, which is the step most stalled programmes skipped. And it governs the estate: owners, review dates, retirement of dead workflows, and reporting on what each journey contributed. The output is a documented system your team can run, not a dependency on us.

The audit takes a fortnight and usually pays for itself in what it switches off. The data foundation and integrations run two to five weeks depending on how many systems hold the customer record. The first journeys go live inside six to ten weeks, starting with the ones that recover revenue fastest rather than the ones that look most impressive. After that it compounds: new journeys, better segmentation, tested content. We plan the sequence so something useful is live in month one instead of a six-month build with nothing to show.

Usually not, and we will say so plainly. Most underperforming programmes are not platform problems — they are data, design and content problems that would follow you to a new tool at considerable cost. Migration is a real project involving journeys, templates, integrations, historical data and deliverability warm-up, so we only recommend it when the current platform genuinely blocks something you need, and we write out the comparison with costs on both sides. Often the better move is to use less of an expensive platform properly, or to drop to a cheaper tier you actually consume.

More of the drafting and less of the deciding than the pitch decks suggest. AI is genuinely useful for producing message variants, summarising a contact's behaviour before a sales call, classifying enquiries and suggesting next steps. It is not a substitute for a data model or a designed journey: Salesforce found 75% of marketers have adopted AI while 84% still admit to generic campaigns, which is what happens when a fast content engine sits on disconnected data. We use AI where a human still approves the output, and we put the effort into the context that makes it worth reading.

Per journey, not per platform. Each one reports entries, completion rate, where people drop out, and the revenue or pipeline it influenced, alongside operational health — error rates, records failing validation, contacts stuck in a stage nobody moves. Where volume allows we run a holdout so the credit argument is settled with evidence rather than opinion. The monthly review reports what moved, what we changed and what is next, and it includes the journeys that underperformed, since retiring a flow is as valuable as launching one.

It is the normal starting point, not a disqualification. We work through it in a fixed order: agree what each field means with the people who use it, deduplicate and normalise, decide which system owns which record, and fill the gaps that journeys genuinely need rather than boiling the ocean. Anything we cannot trust simply does not drive a decision until it is fixed, so the first journeys use fewer signals and still work. This is also where our CRM analytics team earns its place, and the clean-up pays off far beyond marketing.

Only when sales capacity is the constraint. Where it is, scoring should be built from behaviour that historically predicted revenue — specific pages, repeat visits, pricing interest, demo requests, company fit — rather than points invented in a workshop. We validate the model against closed deals, agree thresholds with sales, route and alert on the ones that matter, and review the model quarterly because buying behaviour drifts. If sales can call everybody, scoring is theatre and we will tell you to skip it and spend the time on the follow-up content instead.

We do by default, in your voice, from a documented brief, and we build them as modular blocks so your team can assemble variants without a developer. This is the step that quietly kills most automation programmes: the workflow gets built, the content never arrives, and the whole thing sits in draft. Plenty of clients prefer to write their own and buy the strategy, data work and build from us — that split works well, provided somebody owns the deadline. Our copywriting team keeps the tone consistent with the rest of your marketing.

Not if it is designed with restraint, and that is a deliberate part of the build. Every journey has frequency caps, quiet hours, suppression during active support conversations and an exit condition, and the caps are global so three flows cannot stack messages in one day. Personalisation uses context a customer would expect you to have, not everything technically available. A preference centre lets people choose topic and cadence, which consistently outperforms guessing. Done this way, automation reduces the noise most brands generate — fewer messages, better timed, easier to act on.

Ask what happens in week one: if the answer is not data and audit, expect workflows built on sand. Ask to see a journey map they produced for another client, since the map shows whether they think about customers or about tools. Ask who writes the content. Ask what your team will be able to change without them at the end, and ask for the documentation format. Good marketing automation agencies are comfortable recommending less: fewer journeys done properly, a cheaper platform tier, or a training engagement instead of a retainer when that is genuinely what you need.

Frequently, and it is often the best value. Many teams have a capable operations person who needs the foundation set and the design decided: data model agreed, integrations built, journeys mapped, templates made easy to assemble, reporting connected. We build that, train the team, document it, and stay available for the harder work — a platform upgrade, a migration, a new product line, a quarter where the pipeline stalls. It is a smaller engagement than a full retainer and it keeps day-to-day control with the people closest to your customers.

The audit and build phase is fixed-scope and quoted up front — estate audit, data model, integrations, the first set of journeys, content and reporting — so you approve a number rather than an open retainer. Ongoing work is a monthly fee scoped to how many journeys you want built and maintained, how much content you need, and how many systems we keep in sync. Platform licences, data enrichment and message costs are listed separately rather than folded into one figure, so you can see exactly what you are buying and where it can be reduced.

You keep everything and it keeps running. The platform, the workflows, the templates, the field definitions, the journey maps and the runbook are all in your accounts and written for someone who was not in the room. We hand over with a session for your team, an inventory of every automation and its owner, and a list of what to watch in the first month. We would rather be kept because the work is worth it than because unpicking it would be painful, so nothing is built in a way that traps you.