Skip to content
Faisal Mahmud TusherDigital marketing, SEO & analytics
with Faisal Tusher

Marketing Technology › MarTech Tool Categories

Handling Requests for New Marketing Tools

How to handle requests for new marketing tools: one route in, the questions it asks, triage against tools you already own, trials with real data and decisions.
Guidebeginner12 min readUpdated

Why tool requests need a front door

Stacks grow by request. A team member sees a tool at an event, a colleague recommends one, a vendor offers a free trial, and the quickest route is to sign up. Each decision is reasonable on the day. The result, a year later, is the picture the marketing technology hub warns about in its table on keeping the stack lean: tools nobody can explain, two tools doing the same job for different teams, and a contract that renews with no owner. The tutorial on reviewing marketing tool licences and seats finds those duplicates afterwards, often on company cards. A front door finds them before they are bought. When duplicate tools turn up alongside rebuilt assets and repeated studies, the diagnosis for marketing teams duplicating work traces each one back to the join between the teams that was missing.

The other pages in the tool categories section start later. The choosing and evaluating tools guide begins once a selection is under way, from written requirements. The marketing tool onboarding SOP begins once the decision to buy is made. Neither covers the moment someone says "we need a tool for this", which is where most stacks are actually decided.

The front door works best as a series of gates, each with criteria agreed in advance, the shape the phase-gate reviews framework describes: at every gate a request can go on, change course, wait or stop. For tools, the gates are request, triage, evaluation and onboarding, and most requests should stop at triage with a quick answer.

GateWhat happensWho decidesWhere it is covered
RequestOne route in; a short form with the use case, users, data, owner and budgetThe requester fills it inThis guide
TriageCompared with the tools already owned and the use-case list; one of four outcomesThe person who keeps the stack recordThis guide
EvaluationRequirements, scoring, a proof of concept, total cost and a decision recordThe sponsor and the budget holderThe choosing and evaluating tools guide
OnboardingApproval, privacy review, access, integration and tagging, testing, training, adoption reviewMarketing operations, the tool owner and the data protection leadThe onboarding SOP

One route in and the questions it asks

The route matters more than the form. If requests can arrive by email, in chat, in a meeting and on a card statement, the process sees only the ones people choose to send. Make one route the only route, in whatever system the team already uses for requests; the guide to work management and marketing calendars makes the same rule for marketing work generally, that a request is not accepted until it is in the intake form. The analytics request process is a useful model from a neighbouring team: its intake asks for the decision behind the question, not just the question.

The form should be short enough that nobody avoids it. Five questions do most of the work.

QuestionWhy it is askedWhat a weak answer usually means
What do you need to do that you cannot do now?It turns a product name into a use case the stack can be checked againstThe request is for a product, not a job; ask again
Who will use it, and how often?It sizes the need and shows whether the job is one person's or a team'sAn occasional need that an existing tool or a colleague could meet
What data will it receive or hold?Anything touching personal data needs a review before real data arrivesNobody has thought about it yet; find out before any trial with real data
Who will own it?A tool needs a named owner who can administer it, answer for it and renew itNo owner, which the stack hub treats as evidence the tool should not be in the stack
What budget is there, and from whom?It sets the scale of what is reasonable and names who decidesA cost nobody has agreed to carry

Two further questions are worth asking where they apply: does it need a script on the website, and does it need to connect to other tools? Each one brings a review of its own, covered below, and knowing early keeps the request from stalling later.

Triage: already owned, extend, evaluate or decline

Triage compares the request with what the business already has. The test comes from the designing a marketing stack guide: does the tool serve a written use case, and which? Look the use case up in the stack's use-case list and the stack record. The tutorial on writing the use-case list shows how to build that list and keep it current. Look for tools in the same category, which the MarTech categories guide helps identify when a product describes itself in unfamiliar terms, and for features in tools already owned that the requester may not know about. Then give the request one of four outcomes.

Already covered. A tool the business owns already does the job. The answer is access, training or a pointer to the person who knows how, and it is usually the fastest outcome of all. Where the existing tool does the job badly, record that against the existing tool; it is evidence for its next review.

Extend what exists. An owned tool could do the job with a module, a configuration change or an integration. That is still a change: a new module that handles different data or a new integration is a major upgrade under the onboarding SOP and goes through its procedure. But it adds no new supplier, no new login and no new copy of the customer, and the use case is tested where the data already is. If the stack has a customer data platform (CDP), check whether the need is a segment or an audience the data layer can supply rather than a tool.

Evaluate. No owned tool can reasonably do the job, the use case is real and someone will own the result. The request becomes a selection, with requirements written before any demo, and the choosing guide's method applies; the marketing tool evaluation template holds the record. Agree at this gate what the evaluation must show and by when, and who decides at the end.

Decline. The use case is not one the business needs, the cost is not justified, nobody will own it, or the risks cannot be accepted. Say why, in writing, and say what would change the answer. A declined request with a reason is far more useful than one left unanswered, because the requester stops looking for a way around the process and the log records a need that may return.

Free trials, free tiers and sign-ups on a card

Free tools and trials are where most unreviewed tools enter, because they need no budget approval. The cost of a free tool is rarely the licence. It is the data it receives, the script it puts on the site, the login nobody removes and the habit that becomes a paid subscription.

The rule that matters is about data, not money. The onboarding SOP treats a trial that will receive real customer data as needing its full privacy review, and keeps a trial that uses only dummy data out of scope until it touches real data. The reason is the one the GDPR glossary entry gives: every purpose for processing personal data needs a lawful basis chosen before processing starts, and only the data the purpose needs should be used. Personal data is defined broadly, so an email address, a name or an online identifier is enough to bring a trial into scope. Uploading a customer list to try a tool's features, or connecting a trial to the CRM, is real data. Decisions about whether and how a particular tool may receive it belong to whoever owns data protection; this guide claims nothing about any tool's data handling or terms.

Two more rules keep trials tidy. A trial that needs a script on the website goes through tag governance, through the tag manager (GTM) container with a removal condition, because the tag governance guide notes that a trial script often outlives the trial. And every trial has an owner and an end date in the request log, so that when the date arrives somebody decides whether it becomes a request for evaluation or stops, with its access removed.

Card sign-ups are trials by another name. Where someone has already bought a tool, the request process still applies, after the event: log it, triage it, and if it is kept, take it through onboarding. If it holds real customer data that has not been reviewed, pause its use of that data until the review is done. Finance can help by flagging software on card statements to the person who triages requests.

Requests for AI tools

AI tool requests follow the same front door, with two differences. First, the business may already have a policy that settles which AI tools and uses are approved. The tutorial on writing a marketing AI use policy builds one around a register of uses, and an AI request is triaged against that register as well as against the stack. Second, the "already covered" outcome has a specific form: AI features built into tools the business already runs. The comparison of standalone AI assistants and built-in AI features sets out how to decide which route a task needs.

When an AI request goes to evaluation, the guide to evaluating AI tools for a marketing team adds the questions only AI tools raise, on output quality with the team's own tasks, data and training terms, admin controls and model changes. Those questions belong to that guide and the policy, not to the request form; the form only needs to flag that the request is for AI and what data would go into it.

Turnaround, records and the request log

A front door is judged by how quickly it answers. Decide what turnaround the team can keep for triage, publish it and measure against it; the number depends on the team, but a request that waits too long for triage becomes a card sign-up. Evaluations take longer and should have their own agreed dates, set at the triage gate.

Every request goes in a log, whatever its outcome. The log is a record of decisions and, over time, a picture of what the team keeps needing.

FieldWhat to record
RequestDate, requester, team and the use case in their words
AnswersUsers, data touched, proposed owner, budget, site script, integrations
Triage outcomeAlready covered, extend, evaluate or decline, with the reason
Decision owner and dateWho decided, when, and the date of any next gate
LinksThe evaluation record, the onboarding record or the tool it was pointed to
TrialAny trial running, its owner, its end date and whether it uses real data
Follow-upWhat happened: adopted, extended, bought, declined, or the trial ended

Read the log from time to time as a whole. Repeated requests for the same job show a use case the stack does not serve well, which belongs on the design roadmap. Many requests answered "already covered" show that people do not know what the stack can do, which is a training problem. A run of declined requests from one team may show a need the business has not recognised. The success measure for the tool that is eventually bought, the KPI it is meant to move, starts in the request: the requester's answer to "what do you need to do that you cannot do now?" is its first draft. The marketing stack management playbook gives the log a weekly triage slot and reads it as a whole each quarter, beside the access and seat reviews.

A coach tour operator triages its requests (illustrative)

Illustrative example — not a real result

Sordavex Coach Tours triages a quarter of tool requests

Sordavex Coach Tours, an invented coach tour operator, sets up a request form in its work management system and logs every tool request for a quarter. It receives 11.

Six are already covered. Three people ask for a social scheduling tool when the business already pays for one with unused seats; two ask for a form builder when the email platform builds forms; one asks for a design tool the brand team already licenses. Each gets access and a short handover from the tool's owner, most within 2 working days. Two are met by extending what exists: the booking system's reporting module covers a request for departure reports, and a new integration between the CRM and the email platform covers a request for a separate segmentation tool, each going through the onboarding SOP as an upgrade. One, a request for a tool to manage group booking enquiries from schools, goes to a full evaluation with a sponsor, a budget holder and a decision date 8 weeks out. One is declined because nobody would own it.

The eleventh is a free trial of a review widget that the customer experience team had already started. It turns out to have received the email addresses of 640 past passengers through a list upload. The trial is paused, the data protection lead is brought in to complete the privacy review the onboarding SOP requires, and the widget's script, added to the site directly, is moved into the tag manager under tag governance. By the end of the quarter, the log shows that more than half the requests came from people who did not know what the stack already did, and the team adds a one-page list of owned tools and their owners to its new-starter pack.

Common mistakes

  • Many routes in. Requests by email, chat and card each bypass the process; make one route the only route.
  • A form nobody fills in. Long forms push requests to the side door. Ask the five questions and stop.
  • Triaging on the product, not the use case. Ask what the requester needs to do; the product is one possible answer.
  • Treating free as harmless. A free tool that receives customer data needs the same privacy review as a paid one, before the data arrives.
  • Letting trials run without an end. Every trial has an owner and an end date, and its script and access go when it stops.
  • Restating the AI policy in the request form. Route AI requests to the policy and the evaluation guide; the form only flags them.
  • Declining without a reason. An unexplained no teaches people to use the side door next time.
  • No log. Without it, the same request is answered differently each time and nobody sees the pattern.

Frequently asked questions

Won't a request process just slow everyone down?

Only if it is built as a hurdle. A good front door asks a handful of questions, answers most requests quickly, and often answers them with a tool the team already pays for, which is faster than buying anything. What it slows down is the request that would have put customer data into an unreviewed tool or added a third tool for the same job. Publish how quickly requests are triaged, keep to it, and the route becomes the easiest way to get a tool rather than an obstacle.

Can someone start a free trial while the request is being considered?

Yes, if the trial uses only made-up or dummy data and adds nothing to the website. The onboarding SOP treats a trial that will receive real customer data as needing its full privacy review first, and a trial that needs a script on the site goes through tag governance like any other tag. Free does not change either rule: a free tool that receives customer details is processing personal data, and the business is responsible for that whatever the price.

Who should decide on tool requests?

One named person triages, usually whoever keeps the stack record in marketing operations, because they know what the business already owns. They answer the requests that are already covered or can be met by extending a tool. Requests that go to evaluation have a sponsor and a budget holder who decide at the end of it, using the method in the choosing and evaluating tools guide. Whoever owns data protection is consulted on anything that will touch personal data.

Terms used on this page

Frameworks

Where this page is used

Sources

  1. Regulation (EU) 2016/679 (GDPR) — EUR-Lex

Related